You are on page 1of 121

The University of Maine

DigitalCommons@UMaine
Electronic Theses and Dissertations Fogler Library

Fall 12-18-2015

Design, Manufacturing and Control of an


Advanced High-Precision Robotic System for
Microsurgery
Arezoo Ebrahimi
University of Maine, arezoo_ebrahimi@umit.maine.edu

Follow this and additional works at: http://digitalcommons.library.umaine.edu/etd


Part of the Other Mechanical Engineering Commons

Recommended Citation
Ebrahimi, Arezoo, "Design, Manufacturing and Control of an Advanced High-Precision Robotic System for Microsurgery" (2015).
Electronic Theses and Dissertations. 2378.
http://digitalcommons.library.umaine.edu/etd/2378

This Open-Access Thesis is brought to you for free and open access by DigitalCommons@UMaine. It has been accepted for inclusion in Electronic
Theses and Dissertations by an authorized administrator of DigitalCommons@UMaine.
DESIGN, MANUFACTURING AND CONTROL OF AN ADVANCED

HIGH-PRECISION ROBOTIC SYSTEM

FOR MICROSURGERY

By

Arezoo Ebrahimi

BSIE Sharif University of Technology, 2006

A THESIS

Submitted in Partial Fulfillment of the

Requirements for the Degree of

Master of Science

(in Mechanical Engineering)

The Graduate School

The University of Maine

December 2015

Advisory Committee:

Mohsen Shahinpoor, Professor of Mechanical Engineering, Advisor

Vincent Caccese, Professor of Mechanical Engineering

Xudong Zheng, Assistant Professor of Mechanical Engineering


THESIS ACCEPTANCE STATEMENT

On behalf of the Graduate Committee for Arezoo Ebrahimi I affirm that this

manuscript is the final and accepted thesis. Signatures of all committee members

are on file with the Graduate School at the University of Maine, 42 Stodder Hall,

Orono, Maine.

Dr. Mohsen Shahinpoor, Professor of Mechanical Engineering December 4, 2015

ii
LIBRARY RIGHTS STATEMENT

In presenting this thesis in partial fulfillment of the requirements for an

advanced degree at the University of Maine, I agree that the library shall make it

freely available for inspection. I further agree that permission for "fair use" copying

of this thesis for scholarly purposes may be granted by the Librarian. It is

understood that any copying or publication of this thesis for financial gain shall not

be allowed without my written permission.

Signature:

Date:
DESIGN, MANUFACTURING AND CONTROL OF AN ADVANCED

HIGH-PRECISION ROBOTIC SYSTEM

FOR MICROSURGERY

By Arezoo Ebrahimi

Thesis Advisor: Dr. Mohsen Shahinpoor

An Abstract of the Thesis Presented


in Partial Fulfillment of the Requirements for the
Degree of Master of Science
(in Mechanical Engineering)
December 2015

Microsurgeries like ophthalmic surgery confront many challenges like limited

workspace and hand motion, steady hand movements, manipulating delicate thin

tissues, and holding the instrument in place for a long time. New developments in

robotically-assisted surgery can highly benefits this field and facilitate those

complicated surgeries. Robotic eye surgery can save time, reduce surgical

complications and inspire more delicate surgical procedures that cannot be done

currently by surgeon’s hands.

In this thesis work, the requirements for ophthalmic surgeries were studied and

based on that a robotic system with 6 DOF is proposed and designed. This robotic
system is capable of handling the position and orientation of the surgical instrument

with theoretical accuracy of 10 μm. The design features a remote center of motion

that defines the point of entry into the eye or patient’s body. The forward and

inverse kinematics equations and workspace analysis of the robot is also discussed

and presented.

Six miniature DC motors with their PID controllers were installed on robot arms in

order to run 6 DOF systems. Therefore, the dynamic behavior of a DC motor was

studied and modeled and then the position and velocity transfer functions were

derived and used to study the behavior of the system and also to manually tune the

PID controller.

The function of different elements of the control system including encoder,

controller modules, Controller Area Network (CAN) and the controller software

were discussed as well. The graphical user interface called EPOS Studio and

performs as the motion controller is introduced and the way it organizes

communications among the elements of the control system was described.


ACKNOWLEDGEMENTS

The completion of this undertaking could not have been possible without the

participation and assistance of so many people.

First of all, I would like to express my gratitude to my advisor Prof. Mohsen

Shahinpoor, for his continuous support, patience, motivation and encouragement

throughout my graduate studies.

Besides my advisor, I would like to thank the rest of my thesis committee: Prof.

Vincent Caccese and Dr. Xudong Zheng, for their encouragement and useful

comments.

Most of all, I would like to thank my family for their countless love and endless

support and inspiration.

iii
TABLE OF CONTENTS

ACKNOWLEDGEMENTS ............................................................................................................... iii

LIST OF TABLES ........................................................................................................................... viii

LIST OF FIGURES ............................................................................................................................ ix

CHAPTER 1 : INTRODUCTION .................................................................................................... 1

1.1 Robot Assisted Surgery ..................................................................................................... 2

1.2 Eye Surgery ............................................................................................................................ 4

1.2.1 Cataract Surgery .......................................................................................................... 6

1.2.2 Glaucoma Surgery ....................................................................................................... 7

1.2.3 Refractive Surgery ...................................................................................................... 8

1.2.4 Vitreo-Retinal Surgery .............................................................................................. 8

1.2.5 Eye Muscle Surgery .................................................................................................... 9

1.2.6 Cataract Surgery Procedure .................................................................................... 9

1.3 General Design Requirements for Eye Surgical Robot ....................................... 14

1.3.1 Safety and Accuracy Considerations................................................................. 14

1.3.2 Required Performance and Degrees of Freedom ........................................ 16

1.3.3 Required Velocities and Forces at the Tip ...................................................... 17

1.4 Current Robotic Ophthalmic Surgery Systems ..................................................... 18

1.4.1 Eye Surgery with the da Vinci® ......................................................................... 18

iv
1.4.2 EYE-RHAS (Eye Robot for Haptically Assisted Surgery)........................... 20

1.4.3 University of Tokyo ................................................................................................. 21

1.4.4 JPL RAMS (Robot-Assisted Micro-Surgery) ................................................... 23

1.4.5 UCLA – IRISS............................................................................................................... 24

CHAPTER 2 : DESCRIPTION OF DESIGNED ROBOTIC SYSTEM .................................. 26

2.1 Joints Design and Configuration ................................................................................. 26

2.1.1 Ring Mechanism ....................................................................................................... 27

2.1.2 Double Rotor Mechanism...................................................................................... 28

2.2 Axes of Motion ................................................................................................................... 30

2.3 Motor Selection: Forces, Velocities and Precision ............................................... 32

2.4 Forward Kinematics ........................................................................................................ 36

2.5 Inverse Kinematics .......................................................................................................... 41

2.6 Workspace ........................................................................................................................... 43

2.7 Accuracy ............................................................................................................................... 45

CHAPTER 3 : CONTROL THEORY OF THE SYSTEM......................................................... 47

3.1 Introduction to Theory of Control ............................................................................. 47

3.2 Closed Loop Transfer Function ................................................................................... 48

3.3 Dynamic Model of a DC Motor ..................................................................................... 50

3.4 Motor (V) Transfer Function Analysis...................................................................... 55

3.4.1 Step and Impulse Responses ............................................................................... 56

v
3.4.2 Frequency Domain Analysis ................................................................................ 58

3.5 Proportional Controller ................................................................................................. 59

3.6 PID Controller .................................................................................................................... 63

3.6.1 Transfer Function of PID Controller ................................................................. 65

3.6.2 Proportional Gain Effect ........................................................................................ 66

3.6.3 Integral Gain Effect .................................................................................................. 67

3.6.4 Derivative Gain Effect ............................................................................................. 68

3.6.5 PID Controller Gain Values ................................................................................... 69

CHAPTER 4 : CONTROLLER ARCHITECTURE ................................................................... 70

4.1 Encoder................................................................................................................................. 70

4.2 Controller Module ............................................................................................................ 73

4.2.1 Maxon Controller for EC Brushless Motors ................................................... 75

4.3 Controller Area Network (CAN).................................................................................. 79

4.4 Easy Positioning System (EPOS) Controller Software ....................................... 81

4.4.1 The Motion Controller ............................................................................................ 83

4.4.2 Position Accuracy ..................................................................................................... 87

CHAPTER 5 : CONCLUSION ....................................................................................................... 89

5.1 Conclusions ......................................................................................................................... 89

5.2 Recommendations for Future Work ......................................................................... 91

BIBLIOGRAPHY ............................................................................................................................. 93

vi
APPENDIX A: MATLAB CODE FOR FORWARD KINEMATICS ...................................... 99

APPENDIX B: MATLAB CODE FOR INVERSE KINEMATICS ........................................ 101

APPENDIX C: MATLAB CODE TO PLOT RESPONSES OF THE DC MOTOR SYSTEM

............................................................................................................................................................ 103

BIOGRAPHY OF THE AUTHOR............................................................................................... 105

vii
LIST OF TABLES

Table 2-1: Degrees of freedom. ...................................................................................................... 27

Table 2-2: Actuators requirements............................................................................................... 33

Table 2-3: Selected motor specifications.................................................................................... 34

Table 2-4: Denavit-Hartenberg link parameters ..................................................................... 39

Table 2-5: Misplacement variables ............................................................................................... 46

Table 4-1: DIP switch setting for robotic system with 6 nodes. ........................................ 77

Table 4-2: Data sheet to convert the motion of six axes to quad count (qc) unit ....... 86

viii
LIST OF FIGURES

Figure 1-1: Block diagram for standard master-slave robotic surgical system

(Mahpeykar 2013). ....................................................................................................... 3

Figure 1-2: Anatomy of the human eye (Biographicsmedia 2006) ..................................... 5

Figure 1-3: Anatomy of the eye - anterior and posterior parts. ........................................... 6

Figure 1-4: Cataract eye (left), healthy eye (right) .................................................................... 6

Figure 1-5: Effect of cataract on vision........................................................................................... 7

Figure 1-6: Using speculum to hold eye open........................................................................... 10

Figure 1-7: Making incision (ORBIS Cyber-Sight 2006). ...................................................... 11

Figure 1-8: Viscoelastic injection (ORBIS Cyber-Sight 2006)............................................. 11

Figure 1-9: Creating the main incision (ORBIS Cyber-Sight 2006) .................................. 12

Figure 1-10: Capsulorhexis .............................................................................................................. 12

Figure 1-11: Hydro-dissection (ORBIS Cyber-Sight 2006) and (Robinson 2012) ..... 13

Figure 1-12: Phacoemulsification.................................................................................................. 13

Figure 1-13: IOL insertion. ............................................................................................................... 14

Figure 1-14: Anterior and Posterior manipulation angles (β) ........................................... 16

Figure 1-15: Range of angles that an instrument enters the eye (α) ............................... 17

Figure 1-16: Da Vinci® system for eye surgery....................................................................... 19

Figure 1-17: Eye-RHAS, eye robot for haptically assisted surgery (Meenink

2011) ............................................................................................................................... 21

Figure 1-18: EYE-Rhas, master console (Hendrix 2011) ..................................................... 21

Figure 1-19: Robotic slave system for vitreo-retinal surgery (Ueta, et al. 2009) ....... 22

ix
Figure 1-20: Robotic master-slave system (Ueta, et al. 2009) ........................................... 23

Figure 1-21: JPL Robot-Assisted Micro-Surgery (RAMS) system. .................................... 23

Figure 1-22: UCLA's IRISS robotic eye surgery system. ....................................................... 25

Figure 1-23: IRISS slave robot design. ......................................................................................... 25

Figure 2-1: Ring and overhead crane mechanism .................................................................. 28

Figure 2-2: Double-rotor mechanism .......................................................................................... 29

Figure 2-3: The linear 45° joint. ..................................................................................................... 29

Figure 2-4: Ring and linear motion, Axis-I & II. ....................................................................... 30

Figure 2-5: Double rotor joints, Axis III & IV............................................................................. 31

Figure 2-6: Axis-V and VI, instrument carrier .......................................................................... 32

Figure 2-7: Link parameters: length, twist, offset and angle (Craige 2004) ................. 36

Figure 2-8: Defining joints axes...................................................................................................... 38

Figure 2-9: Link coordinate frames assignment. ..................................................................... 38

Figure 2-10: Reachable workspace of the robot. ..................................................................... 44

Figure 2-11: Double-rotor conic workspace; positive and negative sections. ............. 45

Figure 3-1: A simple input/output system. ............................................................................... 47

Figure 3-2: Unit impulse 𝛿(𝑡) and response to a set of individual impulses. ............... 47

Figure 3-3: Closed loop system. ..................................................................................................... 48

Figure 3-4: A DC motor composition............................................................................................ 50

Figure 3-5: A DC motor modeled as an electric circuit of the armature with

mechanical load. .......................................................................................................... 51

Figure 3-6: A block diagram of the DC motor. .......................................................................... 54

Figure 3-7: The step and impulse responses of the position transfer function........... 56

x
Figure 3-8: The step and impulse responses of the velocity transfer function. .......... 57

Figure 3-9: The position and velocity responses of the system to the voltage

frequency. ...................................................................................................................... 58

Figure 3-10: A simple closed loop system with proportional controller. ...................... 59

Figure 3-11: The proportional controller step response for Kp=1. .................................. 61

Figure 3-12: The proportional controller step response for Kp=5. .................................. 62

Figure 3-13: The proportional controller step response for Kp=10................................. 62

Figure 3-14: The proportional controller step response for Kp=50................................. 63

Figure 3-15: The block diagram of closed loop PID controller. ......................................... 64

Figure 3-16: The PID controller step response for Kp=1 (left) and Kp=50 (right). .... 66

Figure 3-17: The PID controller step response for Kp=500................................................. 66

Figure 3-18: The PID controller step response for Ki=1. ...................................................... 67

Figure 3-19: The PID controller step response for Ki=100. ................................................ 67

Figure 3-20: The PID controller step response for Kd=0 (left) and Kd=1 (right). ....... 68

Figure 3-21: The PID controller step response for Kd=2 ...................................................... 68

Figure 3-22: The best PID response. ............................................................................................ 69

Figure 4-1: The microsurgery robotic system with its motors and controllers

connected through the USB port to the controller software. .................... 70

Figure 4-2: Motor (I) encoder and its cable............................................................................... 71

Figure 4-3: The signals of an incremental encoder. ............................................................... 72

Figure 4-4: Physical layout of the six motor controllers CAN bus. ................................... 74

Figure 4-5: The EC brushless controller, interfaces and designations ........................... 75

Figure 4-6: JP1 – Numbering scheme........................................................................................... 76

xi
Figure 4-7: Controller (III) in operating status with the Green LED “ON”. ................... 78

Figure 4-8: DC brushed controller, interfaces and designations....................................... 78

Figure 4-9: Motor/encoder connector (left) and DC brushed controllers (V) &

(VI) wired to the system (right). ........................................................................... 79

Figure 4-10: Schematic overview of the controller with external connections

and communication. .................................................................................................. 81

Figure 4-11: The EPOS Studio screen with the Workspace tab open. ............................. 82

Figure 4-12: Communication set up with six nodes. .............................................................. 83

Figure 4-13: Different operating modes under the “Tools tab”. ........................................ 84

Figure 4-14: Data entry for 45 degree motion of node 1 (around the ring) in

profile position mode. ............................................................................................... 85

xii
CHAPTER 1 : INTRODUCTION

Robotics is a rapidly growing field that has applications in diverse industries

including manufacturing, automated production lines, welding, painting, spraying,

assembly, packaging, product inspection, testing and many more. The advantages of

employing robotic systems are due to their ability to produce more accurate and

high quality work in a short amount of time which significantly reduces the cost of

production and makes them the best fit to perform many industrial tasks.

A robot is a mechanical or artificial executer, usually an electro-mechanical

machine that is guided by a computer program or electronic circuit which can be

autonomous or semi-autonomous. Also there are some tasks that require such high

level of accuracy and precision that cannot be achieved using human hands. This is

one of the most promising attributes of robots and based on this feature medical

application of robots is recently in demand.

Designing a robot for medical applications highly benefits hospitals, medicine,

patients and doctors. It requires versatility and a specific design for certain

applications. Medical robots can assist in two different areas of health care and

surgeries.

Health care robots are designed to perform a variety of tasks to help patients,

elderly or individuals while surgical robots are designed to assist surgeons perform

very specific tasks with high accuracy and less trauma. The objective of these robots

is to enhance surgical skills and to help surgeons to perform complicated

1
procedures more ergonomically and non-invasively with more precision and fewer

efforts.

1.1 Robot Assisted Surgery

One of the fields that highly benefits from robotic advances is robotic surgery.

Surgery performed by surgeons is quickly moving towards less invasive and more

precise surgery to reduce hospitalization time and expedite recovery and healing.

Although improvement on manual procedures has yielded successively better

results, these are merely marginal and do not reflect the real needs for effective

techniques (Gheshmi 2012). Therefore, the need for utilizing new techniques in

performing surgeries can be seen through robotic surgery (Patel 2008). There are

some surgical procedures that cannot be done by hand or they are extremely

difficult to perform, such as delicate eye surgeries or brain surgery in which tiny

blood vessels are blocked or aneurysm or hemorrhage have occurred. Currently,

surgeons practically use other indirect methods for those kinds of surgeries that

cause more trauma and increase healing and hospitalization time and cost in some

cases.

Robotic surgery is already a possibility for robotic microsurgery (Shahinpoor

and Gheshmi 2014). There have been recent successes in endoscopic and

laparoscopic techniques, visualization and surgical instruments which have enabled

performing a number of traditionally manual surgeries through robotic surgery in

less time and with less pain. For example, in laparoscopy, large incisions have been

substituted by a few small incisions (up to five, 5-12mm in diameter) and therefore

2
there’s no need to cut the body open anymore to reach inside the abdominal area.

The robotic arms are equipped with end-effectors having grippers, haptic, optical

and force sensors and other surgical tools which perform their functions and

continuously transfer information to the surgeon.

The term “robotically-assisted surgery” is often used in literature (Ghani, et al.

2012) (Lambaudie, et al. 2010) to emphasize that the robot is fully under the control

of the surgeon. A master-slave system is the common layout used in robotic assisted

surgery which consists of master console, controller and slave robot. The slave

system manipulates the surgical instruments and tools, while the surgeon does the

surgery by controlling the master console. Figure 1-1 shows a block diagram of this

type of master-slave surgical systems layout.

Figure 1-1: Block diagram for standard master-slave robotic surgical system

(Mahpeykar 2013).

3
Tele-operation is one of the spectacular advantages of this system, enabling the

surgeon to perform the surgery from a remote location. One of the commercially

available surgical master-slave systems is the Da Vinci® System by (Intuitive

Surgical 2013). Performing surgery using this system significantly reduces the

trauma because of using long slender instruments (300mm long, 8.5 mm in

diameter) that enter the human body via small incisions, often fitted with a trocar.

These instruments are manipulated outside the body and their movements

inside the body are inverted and scaled by the instrument pivoting point (at the

insertion site).

One of the other benefits of using a master-slave layout is scaling the surgeon’s

hand motion and filtering the tremor which will reduce the probability of unwanted

collision between surgical tools with adjacent tissues in high precision surgeries and

therefore reduces human error possibilities.

1.2 Eye Surgery

Eye surgery, also known as ocular surgery, is surgery performed on the eye or its

adnexa, typically by an ophthalmologist (Encyclopedia of Surgery 2015). The eye is

a fragile organ, and requires extreme care before, during, and after a surgical

procedure. The anatomy of human eye is shown in Figure 1-2. An expert eye

surgeon is responsible for selecting the appropriate surgical procedure for the

patient, and for taking the necessary safety precautions.

Robot-assisted methods offer many advantages over conventional surgical

techniques such as improved precision, reduced tremor, and amplified scale of

4
motion. One of the eye surgeries performed on the inner side of the eyeball is

refractive cataract surgery. The other type is external eye surgery which is

performed on the rectus muscles controlling the eye movement. Refractive surgery

is often external, only affecting the cornea (LASIK 1 for example) or can be

intraocular. Figure 1-3 shows the anterior and posterior sections of the eye. The

majority of eye surgeries, almost 80% of cases, are performed on the anterior part

of the eye.

Figure 1-2: Anatomy of the human eye (Biographicsmedia 2006)

1 Laser-assisted in situ keratomileusis


5
Figure 1-3: Anatomy of the eye - anterior and posterior parts.

1.2.1 Cataract Surgery

When the eye’s naturally clear lens becomes opaque or clouded, it’s called a

cataract. Most cataracts are the result of the natural process of aging. Others may be

present at birth or develop as a result of physical, drug, or chemical injury. Cataract

surgery, one of the most common operations performed in the U.S., clears up the

cloudiness. Figure 1-4 shows a cataractous eye lens compared to normal clear lens.

Figure 1-4: Cataract eye (left), healthy eye (right)

6
Figure 1-5: Effect of cataract on vision

Cataracts could be treated using eye drops in the early stages of development.

However, surgical treatment is required if the cataract is fully developed. During

cataract surgery, the cloudy lens is removed or cleaned out and replaced by a clear

manmade lens called artificial intraocular lens (IOL). With over 1.8 million surgeries

a year in the US, cataract surgery is the most common eye surgery in the United

States (Goroll and Mulley 2012).

1.2.2 Glaucoma Surgery

The glacoma disorder happens due to damage of optic nerve which is usually

resulted from high pressure in the eye due to the aqueous humor. The reason of

high pressure is unbalanced production and drainage of aqueous fluid inside the eye

globe. Unfortunately the damage is permanent however there are some surgical

7
treatments to control the situation and prevent the eye from further damages. These

surgical treatments involve with opening the blocked schlemm's canal and

trabecular mesh tissues to let the aqueous humor fluid drain properly and reduce

the intraocular pressure (IOP).

1.2.3 Refractive Surgery

A refractive problem is an issue of eye lens to focus parallel rays of light properly

to produce a clear image on the retina. Any defect in the shape or curvature of the

cornea, length of the eyeball or age of the lens can cause refractive errors. This

problem can be temporarily corrected using eye glasses or contact lenses. To treat

this disorder permanently refractive surgery known as LASIK is performed. In this

surgical treatment the cornea is cut open to access the eye lens and then an excimer

laser is used to treat the surface to the desired shape.

1.2.4 Vitreo-Retinal Surgery

Vitreo-retinal surgery refers to any ophthalmic surgery performed to treat

problems related to the retina such as retinal detachment, vitrectomy and diabetic

retinopathy.

The retina is responsible to generate electrical impulse of image and sends this

information via the optic nerve to the brain. This image is created from focused light

rays entering the eye through the cornea and the pupil. Even though light is needed

to generate an image, it also can damage retinal structures when too much of it is

focused onto the retina.

8
To fix the damaged retina, the vitrectomy surgery is performed by making three

small incisions through sclera. The surgery is performed through an operating

microscope. For more extensive work, other instruments such as microscissors or

microforceps may also be introduced through these incisions to perform

microsurgery near the retinal surface.

1.2.5 Eye Muscle Surgery

Extraocular muscle surgery is the third most common eye surgery in the United

States (American Society of PeriAnesthesia Nurses 2012). Eye muscle surgery treat

diseases related to the alignment of eyes such as strabismus. Eye muscles are

strengthened, weakened or repositioned during this operation. To perform this, a

small incision is made in the thin white tissue (conjunctiva) overlying the muscle.

The muscle is then separated from the eye and reattached in a new position.

Ophthalmic surgery is not only limited to mentioned categories and

includes more types such as oculoplastic surgery, corneal surgery, eye removal,

and so on. In the next section, cataract surgery is discussed in detail as it is one of

the target surgeries of the intended robotic system design. Design requirements are

mostly derived from the procedures in cataract surgery as they are applicable to

other ophthalmic surgeries as well.

1.2.6 Cataract Surgery Procedure

In general cataract surgery procedure involves with replacing the cataractous

natural lens with an artificial intraocular lens (IOL). The most popular technique to

9
perform this surgery is Phacoemulsification. In this method an ultrasonic needle is

used to emulsify the lens and extract out the particles through small incisions. Also

there are other methods that remove the entire natural lens in one or few pieces

which require bigger incisions that takes longer to heal (Mahpeykar 2013).

The phacoemulsification surgical procedure steps are as follows:

1) Anesthesia and Dilation:

To numb eye nerves and prevent blinking or eye movement during

surgery a topical anesthetic eye drop is used followed by an injection.

This injection dilates the pupil to provide a better access to eye lens.

2) Exposure of the eye ball:

The eye lids are held open using a speculum to prevent blinking

during the surgery, Figure 1-6.

Figure 1-6: Using speculum to hold eye open.

3) Making incision:

A small incision (almost 1mm) is made to provide an entry to the

anterior chamber of the eye for anesthetics and viscoelastic gel

injection to the anterior chamber. It is also used to introduce a second

instrument in the eye later.

10
Figure 1-7: Making incision (ORBIS Cyber-Sight 2006).

4) Injection of viscoelastic gel

To stabilize and maintain the shape of the cornea and also pressurize

the anterior chamber the viscoelastic gel is injected to the anterior

chamber through the small incision made in previous step.

Figure 1-8: Viscoelastic injection (ORBIS Cyber-Sight 2006).

5) Main incision:

To introduce the main instrument, a 2 mm incision is made 90 degree

from the previous smaller incision using diamond knife, shown in

Figure 1-9.

11
Figure 1-9: Creating the main incision (ORBIS Cyber-Sight 2006)

6) Capsulorhexis

In this step a portion of the lens capsule is cut open to access the lens.

To create a round and smooth opening so that the lens can be

emulsified the anterior capsule wall is scratched using a curved

needle and then performing a curvilinear tearing using forceps shown

in Figure 1-10.

Figure 1-10: Capsulorhexis

7) Hydrodissection

To weaken the capsular-cortical connection so that the lens can be

mobile, a cannula is filled with balanced salt solution (BSS) and is

inserted through the small incision and injects a jet of BSS behind the

lens, Figure 1-11. This is verified by rotating the lens in place to make

sure it has been disconnected from the posterior capsule wall.

12
Figure 1-11: Hydro-dissection (ORBIS Cyber-Sight 2006) and (Robinson 2012)

8) Ultrasonic emulsification and irrigation aspiration

In this step a vibrating ultrasonic instrument with titanium or steel tip

is inserted into the anterior chamber of the eye through the main

incision. The cataractous lens is sculpted and emulsified by the tip of

instrument while its particles aspirates through the tip, Figure 1-12.

Figure 1-12: Phacoemulsification.

After clearing all the remnants additional viscoelastic gel is injected to

fill and inflate the empty capsule.

9) Insertion of new lens

Now that the capsule is clean, a folded IOL, as shown in Figure 1-13, is

inserted into the lens capsule via the main incision and then the IOL is

unwrapped. This is also shown in Figure 1-13.

13
Figure 1-13: IOL insertion.

10)Viscoelastic removal

After the implantation of IOL, the viscoelastic gel is removed from the

anterior chamber using an irrigation-aspiration device.

11) Wound sealing and pressure check

As the final step the surgeon seals the incisions by injecting some BSS

to hydrate the incisions so that they do not leak and the pressure of

the eye is also checked to make sure everything is okay.

1.3 General Design Requirements for Eye Surgical Robot

1.3.1 Safety and Accuracy Considerations

The designed robotic system must be able to deliver the required tasks while

following the applicable safety requirements. Safety criteria main considerations

mentioned are as follows (Taylor, et al. 1991). The robot must be fully under control

and the surgeon must be in charge at all times. The instrument tip should stay

14
within a pre-specified positional envelop and never exert excessive force on the

patient.

Also, the robot must be either allowed to fail to a safe state or it must continue to

operate even in the presence of failure (Kazanzides 2009).

In order to provide these criteria the followings safety design requirements

considered:

1) The instruments must be back-drivable.

2) To prevent unnecessary force to be applied on the patient in case of system

malfunction, the active instrument must be mechanically leashed, meaning

that the degrees of freedom manipulating the orientation and position of the

instrument tip need to be rigid and consistent. This also leads to a minimally

invasive surgery.

3) Motion of the surgical instrument must be very steady and accurate with

virtually no shake or vibrations.

4) All joints need to hold their positions in case of power loss.

Eye surgery involves very thin organs. The thickness of the retinal membrane

is only a few microns. The vessels attached to the retina are 50-150 μm in

diameter. Therefore the accuracy of the robot must be less than 50 μm.

Meaning, if the calculated accuracy of the system is above 50 μm, the robot is

not suitable to perform such delicate procedures.

15
1.3.2 Required Performance and Degrees of Freedom

This surgical system is considered to be suitable to perform both anterior and

posterior surgical procedures. For anterior surgical procedures the position of the

instrument must be tangent to the cornea chamber, while in posterior surgical tasks

it is more towards the inside of the eye as shown in Figure 1-14. Here the angle β is

introduced as the angle between the tool axis and the optic axis of the eye and will

be used to model the design.

Figure 1-14: Anterior and Posterior manipulation angles (β)

The other performance criteria is the surgical instrument range of angles

required in different eye surgery tasks like injections, capsulorhexis, vitrectomy,

retinal manipulation and so on. This angle which is here called α is shown in

Figure 1-15.

16
Figure 1-15: Range of angles that an instrument enters the eye (α)

Based on the procedure requirement, the range for α angle is -45°-315°and for β

angle is 0°-90°.

To define the degrees of freedom for this robotic system different ophthalmic

surgeries considered. For example, vitreo-retinal surgery demands 4 degrees of

freedom (Meenink 2011).

For this device to be capable to perform different eye surgeries the robot must

have 6 active degrees of freedom to control the position and orientation of the

surgical instrument. Two of these are to relocate the entry point to the eye; two are

required to orient the tool about this entry point; one is required to slide the

instrument in and out of the eye; and one is required to rotate the tool about its axis.

1.3.3 Required Velocities and Forces at the Tip

The required velocity is usually low for microsurgeries to keep the accuracy

high. Among different ophthalmic surgeries the maximum movement occurs in

17
cataract surgery during capsulorhexis step when the tip of the instruments move 10

mm in 0.4 second. Therefore, a maximum velocity of 25 mm/s is desired at the tip of

the instrument.

To measure the forces required for eye surgery, vitreo-retinal surgery is

considered since required higher forces compared to other types. This surgery

involves with membrane peeling, vessel puncture and vessel dissection. An

advanced instrument equipped with a tri-axial force sensor (X, Y, Z) was used

(Jagtap and Riviere 2004) to measure required force for three mentioned

procedures. On average their respective force in X/Y direction was: 55 mN, 25 mN

and 20mN. In Z direction this was: 375 mN, 75 mN and 575 mN (RMS).

Furthermore, in Z direction, the force range measured: 3140 mN for membrane

peeling, 490 mN for vessel puncture and 5870 mN for vessel dissection (Meenink

2011).

1.4 Current Robotic Ophthalmic Surgery Systems

Robotic eye surgery is particularly attractive due to its advantages such as

precision, accuracy, stability, amplified scale of motion, multitasking, automation,

association of imaging systems, reduced tremor and tele-operation. More recent

robotic eye surgery studies are briefly discussed below.

1.4.1 Eye Surgery with the da Vinci®

18
Da Vinci Surgical System (Intuitive Surgical, Sunnyvale, CA) is the most well-

known platform in robotic surgery. A feasibility study of eye surgery with the da

Vinci® system was done by (Bourla, et al. 2008). In this study, modified robotic

instruments used to perform different intraocular surgeries such as vitrectomy,

removing intraocular foreign body, and anterior capsulorhexis with the da Vinci

system on porcine eyes. In this study the dexterity, maneuverability, and

visualization necessary for intraocular surgery were examined.

Figure 1-16: Da Vinci® system for eye surgery.

Instruments are adapted for the da Vinci® by getting them stuck onto a metal

plate, Figure 1-16 (Pitcher, et al. 2012). The plates are grasped by standard robotic

forceps. 3D endoscope of the da Vinci® provides visual feedback.

This study shows robotic instruments gives a full range of movement. The

dexterity of the robotic arms is excellent, with steady instrument motion. However,

controlling the robotic arms was not as intuitive as moving the wrist. A high stable

point of rotation results in motion-related stress at the site of instrument insertion.

Visualization of the external operative field during intraocular procedures required

19
camera realignment, which here inferior visual feedback compared with an

ophthalmic microscope made anterior segment surgery hard to perform.

1.4.2 EYE-RHAS (Eye Robot for Haptically Assisted Surgery)

A master-slave system for vitreo-retinal eye surgery was developed at

Eindhoven University of Technology (TU/e) by (Meenink 2011) and (Hendrix

2011). This is a two arm robotic system with 4 degrees of freedom each arm and a

fixed remote center of motion and an additional degree of freedom to actuate the

gripper motion of the instrument. The arms can be mounted on the patient’s table.

The master console as shown in Figure 1-18 is two joysticks controls by the

surgeon. The surgeon’s movements scales down and tremor filters out the by the

controller. This system is not capable to relocate the entry point to the eye during

the surgery and therefore is not suitable for cataract surgery.

20
Figure 1-17: Eye-RHAS, eye robot for haptically assisted surgery (Meenink 2011)

Figure 1-18: EYE-Rhas, master console (Hendrix 2011)

1.4.3 University of Tokyo

Another slave robot is developed at University of Tokyo for vitreo-retinal

surgery (Ueta, et al. 2009). This robot has five degree of freedom, shown in Figure 1-

21
19. The system uses master console with a generic seven degree of freedom

featuring 3D vision system, shown in Figure 1-20. Instrument axis has a motion

range of ±50° rotation and -29 to 32 mm axial penetration, and 360° for rotation

about tool’s axis is realized with this slave robot.

The system features a fixed remote center of motion. The accuracy for this slave

system is 5-10 times more than manually performed surgery.

Figure 1-19: Robotic slave system for vitreo-retinal surgery (Ueta, et al. 2009)

22
Figure 1-20: Robotic master-slave system (Ueta, et al. 2009)

1.4.4 JPL RAMS (Robot-Assisted Micro-Surgery)

A cable-driven telerobotic (master-slave) platform for eye surgery was

introduced at NASA’s Jet Propulsion Lab, California (Charles, et al. 1997). To

improve precision and amplify force feedback the system provides scale-down

human-input motions and tremor filtering.

Figure 1-21: JPL Robot-Assisted Micro-Surgery (RAMS) system.

23
The slave robot has two compact arms with 2.5 cm diameter and 6 DOF, shown

Figure 1-21. All joints are cable-driven and are connected to a cylindrical ø120 mm

base at the top that contains the motors. The accuracy of surgical instrument is 12

μm (Das, et al. 1999).

The master arms are very similar to the slave arms and have the same

dimensions with lower gear ratios to provide back-drivability. According to (Pitcher,

et al. 2012) the complexity of the control software and the lack of mechanical

remote center of motion were the main limitations of this system.

1.4.5 UCLA – IRISS

The Intraocular Robotic Interventional Surgical System (IRISS) is the result of

research collaboration between Jules Stein Eye Institute and the UCLA Department

of Mechanical and Aerospace Engineering. This system is designed as a

microsurgical platform capable of performing complete ophthalmic procedures

(Rahimy, et al. 2013). The IRISS system features a head-mounted “True Vision”

(True Vision Displays, Inc., Cerritos, CA) stereoscopic visualization system, two

joystick controls with tremor filtration and scaled motion, appropriately sized

custom designed arms to accommodate different surgical instruments. The IRISS

has the unique capability to manipulate two surgical instruments simultaneously

through ocular incisions spaced millimeters apart.

24
Figure 1-22: UCLA's IRISS robotic eye surgery system.

The slave manipulator, shown in Figure 1-22, can hold two surgical instruments

simultaneously. Two motors drive the curved guides on a XY bed and have

perpendicular axes with the curved guides. The instrument is then mounted on a

carrier that slides over the curved guide. The carrier provides the axial motion of

the instrument to the eye by means of a linear actuator.

Figure 1-23: IRISS slave robot design.

25
CHAPTER 2 : DESCRIPTION OF DESIGNED ROBOTIC SYSTEM

To design the surgical instrument for eye surgery, the requirements discussed in

Chapter 1 are considered. Based on those requirements, the instrument manipulator

must have 6 degrees of freedom which two of them are reserved to relocate the eye

entry point. The first one must provide 45° to 315° peripheral manipulation range

and the second one is to provide 0° to 90° vertical manipulation range. The entry

point to the eye must be fixed, while the surgical instrument must have at least 4

degrees of freedom about the entry point. Also the instrument must be capable of 30

mm penetration inside the eye and exerting push and pull forces up to 5.8 N. The

design must be compact and easy to install and use with accuracy of less than 50 μm

(Mahpeykar 2013).

2.1 Joints Design and Configuration

In this design the entry point is mechanically fixed. The advantages of this

method over any other method which fixes the entry point by software is its higher

accuracy and safety since it requires less actuation and is less influenced by

calculation errors (Ida, et al. 2012).

This approach uses the Remote Center of Motion (RCM) concept. This RCM point

is the intersection point for all degree of freedom axes as well as the entry point to

the eye. These degrees of freedom are listed in Table 2-1:

26
Table 2-1: Degrees of freedom.

Axis DOF Description

1 φ Ring motion

2 R Overhead linear joint

3 θ1 Double-rotor primary axis

4 θ2 Double-rotor secondary axis

5 Z Instrument insertion from entry point to the eye

6 ψ Rotation of the tool about its own axis

To relocate the entry point a 2 DOF planar mechanism is required. Since the

instruments enter the eye from the periphery, a polar ring system seems to be the

best choice. This mechanism can control an angle and a radius to reach to a point.

The mechanism chosen to provide 4 degrees of freedom from the entry point is

the double rotor mechanism due to its simplicity, slender design and low number of

moving parts. The configuration and design of ring and double rotor mechanisms is

described in next sub-sections.

2.1.1 Ring Mechanism

Ring mechanism is designed to generate a rotary motion about the eye and its

350 mm diameter provides required workspace for the optical devices above the

patient’s head. To manipulate the radial position, a linear joint is used. The

configuration is shown in Figure 2-1.

27
The ring itself serves as the surgical head-master and holds the system together.

The linear joint acts like an overhead crane positioning the double-rotor RCM in the

radial direction. These two in combination provide a rigid platform for controlling

the RCM.

Figure 2-1: Ring and overhead crane mechanism

A carriage assembly featuring V-groove bearings is then mounted on the ring

slide. The carriage has a motor driven pinion that is meshed with the external gear

teeth of the ring slide.

2.1.2 Double Rotor Mechanism

The double-rotor mechanism consists of two rotors and three axes holding

constant angle of 22.5° between them by the designed linkages as seen in Figure 2-2.

This is a right-angle conic mechanism which cone’s apex is the RCM point.

28
Figure 2-2: Double-rotor mechanism

Here, the rotations about the primary and secondary axes of the double-rotor

mechanism are called θ1 and θ2 respectively. A linear joint with a 45° angle with the

ring axis connects the double-rotor mechanism to the ring.

Figure 2-3: The linear 45° joint.

29
2.2 Axes of Motion

The ring motion is defined as Axis-I of this robotic system with range of 315°

since the other 45° of the ring is used to mount it on place. The lead screw linear

joint bolted to the bottom of the ring forms the Axis-II of motion. Precision acme

lead screw with anti-backlash nut is used to provide a rigid positioning mechanism.

See Figure 2-4. This lead screw mechanism accommodates a range of linear motion

up to 90 mm.

Figure 2-4: Ring and linear motion, Axis-I & II.

The 45°-link which is installed to the anti-backlash nut of the linear joint from

one side and is connected to the double-rotor mechanism to the other side forms the

Axis-III of the system which is also the primary axis of the double rotor.

The secondary axis of the double rotor is Axis-IV of the system and has the same

design as Axis-III. The first and secondary axes of the double rotor both

accommodate 360° rotation, see Figure 2-5.

30
Figure 2-5: Double rotor joints, Axis III & IV.

A linear joint connected to the end of double rotor mechanism forms Axis-V.

This axis is parallel to surgical tool’s axis and has basically the same design as Axis-

II. To position the surgical instrument a custom made nut with two linear bearings

on the sides and an anti-backlash nut in the middle is used. This linear joint

accommodates 60 mm penetration for surgical instrument. Axis-6 is the axis of

instrument rotation about itself which provides a range of 360° rotation of surgical

device. This is shown in Figure 2-6.

31
Figure 2-6: Axis-V and VI, instrument carrier

In this design motors are direct drive and placed on the joints to eliminate

unnecessary backlash in the system and delivers more efficiency and accuracy.

However, this increases the moving mass of the system and causes some wiring

complications but since light miniature motors are readily available and can be used

here, the benefits overweight the drawbacks.

2.3 Motor Selection: Forces, Velocities and Precision

To properly select motors for this design required force and velocity for each

axis of the robot listed in Table 2-2 base on simple calculation and estimations

considering the weight of links and surgical forces (Mahpeykar 2013).

The selected motor specifications are listed in Table 2-3.

32
Table 2-2: Actuators requirements.

Max. Continuous
Joint Max. Cont. Velocity Cont. Power 100% Desired Position
Mechanism and intermittent
Type Required drive efficiency Accuracy
Force/Torque

Ring 230 N.mm


Axis I Ring Gear/Pinion Drive 540 rpm 12.65 W 3°
Motion 400 N.mm

7N
Axis II Linear Precision Lead Screw 25 mm/s 0.5 W 50 μm
20 N

200 N.mm
Axis III Rotary Direct Geared motor 100 rpm 2W 0.5°
320 N.mm

33
80 N.mm
Axis IV Rotary Direct Geared motor 100 rpm 0.8 W 0.5°
130 N.mm

5N
Axis V Linear Precision Lead Screw 25 mm/s 0.2 W 50 μm
8.5 N

20 N.mm
Axis VI Rotary Direct Geared motor 200 rpm 0.5 W 1.5°
35 N.mm
Table 2-3: Selected motor specifications.

Motor Shaft Diameter Power Voltage Encoder Pulse Gear box

Axis I 22 mm 25 W 24 V 512 19

Axis II 22 mm 25 W 24 V 512 1

Axis III 16 mm 8W 12 V 512 84

Axis IV 16 mm 8W 12 V 512 19

Axis V 10 mm 2.7 W 12 V 512 1

Axis VI 10 mm 2.7 W 12 V 128 16

This system has two linear joints with lead screw and anti-backlash nut design to

provide required positioning accuracy. For the Axis II, a ¼’’ screw with 1.27 mm

lead (0.05”) is used. Therefore, the linear precision of Axis II can be calculated as

follows:

1.27 𝑚𝑚/𝑡𝑢𝑟𝑛 𝜇
𝐴𝑥𝑖𝑠 𝐼𝐼 𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 = × 1000 = 𝟐. 𝟒𝟖 𝝁𝒎
512 𝐶𝑃𝑇 𝑚𝑚

For Axis V, a 9� "


64 screw with lead of 1.22 mm (0.048”) is chosen. Then the

precision would be:

1.22 𝑚𝑚/𝑡𝑢𝑟𝑛 𝜇
𝐴𝑥𝑖𝑠 𝑉 𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 = × 1000 = 𝟐. 𝟑𝟖 𝝁𝒎
512 𝐶𝑃𝑇 𝑚𝑚

34
For Axis I, a pinion used to mesh with the ring gear has 19 teeth and the ring gear

has 456 teeth and a ratio of 1:24 is between actuator shaft and the ring motion. The

angular precision, assuming zero-backlash, would be:

1 𝑒𝑛𝑐𝑜𝑑𝑒𝑟 1 𝑚𝑜𝑡𝑜𝑟 1 𝑃𝑖𝑛𝑖𝑜𝑛 360°


𝐴𝑥𝑖𝑠 𝐼 𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 = � �× � �× � �× = 𝟎. 𝟎𝟎𝟏°
512 𝑚𝑜𝑡𝑜𝑟 19 𝑔𝑒𝑎𝑟ℎ𝑒𝑎𝑑 24 𝑅𝑖𝑛𝑔 1

For Axis III, the angular precision would be:

1 1 360°
𝐴𝑥𝑖𝑠 𝐼𝐼𝐼 𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 = × × = 𝟎. 𝟎𝟎𝟖°
512 84 1

The same configuration with a 19:1 gearhead instead, is used for Axis IV. Therefore

the precision would be:

1 1 360°
𝐴𝑥𝑖𝑠 𝐼𝑉 𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 = × × = 𝟎. 𝟎𝟑𝟕°
512 19 1

For Axis VI, the angular position precision would be:

1 1 360°
𝐴𝑥𝑖𝑠 𝑉𝐼 𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 = × × = 𝟎. 𝟏𝟕°
128 16 1

Brushless motors deliver smooth operation, back drivability, compact design

and high power-to-size ratio and torque.

35
2.4 Forward Kinematics

Now that all the link frames at each joint are defined, kinematic relations between

joints can be describe.

To study forward kinematics of this robotic system, different general methods

can be adopted. One common method is called Denavit–Hartenberg method (D-H). In

this method the geometry of each connection (link) in the mechanism is defined by

4 parameters namely: length, twist, offset and angle. Link parameters for a general

connection are illustrated in Figure 2-7. After assigning coordinate frames to each

link these parameters can be defined as (Mahpeykar 2013):

ai-1 (length): the distance from Zi-1 axis to Zi axis

αi-1 (twist): the angle from Zi-1 axis to Zi axis measured about Xi axis

di (offset): the distance between Xi and Xi-1 measured along Zi axis

θi (angle): the angle from Xi-1 to Xi measured about Zi axis.

The transformation matrices can be found using these parameters.

Figure 2-7: Link parameters: length, twist, offset and angle (Craige 2004)

36
A 4x4 matrix will be resulted from forward kinematics analysis. This matrix

describes the position and orientation of the {nth} link frame with respect to the base

frame {0} and is a function of joint variables (θi), as presented in Eq. 2-1.

Coordinates of the end-effector’s X-axis, Y-axis and Z-axis are described in first three

columns which together describe the orientation of tool frame relative to base

frame. The position of the end-effector’s frame is described in fourth column.

𝒓𝟏𝟏 𝒓𝟏𝟐 𝒓𝟏𝟑 𝒑𝒙


0 𝒓𝟐𝟏 𝒓𝟐𝟐 𝒓𝟐𝟑 𝒑𝒚
𝑛𝑇4×4 (𝜃1 , 𝜃2 , 𝜃3 , … , 𝜃𝑛 ) = �𝒓 𝒓𝟑𝟐 𝒓𝟑𝟑 𝒑𝒛
� Eq. 2-1
𝟑𝟏
0 0 0 1

In next step joint’s axes need to be defined. Z1 represents the axis of the ring

motion corresponding to φ angle. Z2 is the axis of the first linear joint, R position,

corresponding to the radial movement of the RCM. Z3 is the axis of the first rotary

joint of the double-rotor mechanism, θ1 angle. Z4 is the axis of the second rotary

joint of the double-rotor mechanism, θ2 angle. Z5 is the axis of the second linear

joint, corresponding to the linear motion of the surgical instrument, Z position. Z6 is

the tool axis and rotation about itself, ψ angle. These joint’s axes are shown in

Figure 2-8.

Then coordinate frames must be assigned to each link, same way shown in

Figure 2-7. These coordinate frames are attached to the robot links as illustrated in

Figure 2-9, with the frames of the double-rotor mechansim shown in detail on the

right side of the figure.

37
Figure 2-8: Defining joints axes

Figure 2-9: Link coordinate frames assignment.

38
Link parameters are identified and listed in Table 2-4. The ring motion angle is

shown as – 𝝋 since the positive direction for 𝝋 is chosen to be clockwise looking

down from the top of the ring.

Table 2-4: Denavit-Hartenberg link parameters


{𝒊} 𝑎𝑖−1 𝛼𝑖−1 𝑑𝑖 𝜃𝑖
1 0 0° H −𝝋
2 0 90° −𝑹 0
3 0 -135° −𝐻√2 𝜽𝟏
4 0 22.5° 0 𝜽𝟐
5 0 22.5° 𝒁 0
6 0 0° 0 𝝍

All distances between axes 𝑎𝑖−1 are zero in Table 2-4, which means that each

axis intersects with its neighboring axis. This in fact simplifies the kinematic

equations. H is defined as the vertical distance between the base frame and the ring

frame with the following relation applies:

𝑅 − 𝐻√2 = 𝑟 Eq. 2-2

Where, r is the distance between the RCM and the base frame and is solely

controlled by the first linear actuator.

The general transformation matrix from link 𝑖 to 𝑖 − 1 with link parameters

𝑎𝑖−1 , 𝛼𝑖−1 , 𝑑𝑖 and 𝜃𝑖 is presented in Eq. 2-3 below:

𝑐𝜃𝑖 −𝑠𝜃𝑖 0 𝑎𝑖−1


𝑖−1 𝑠𝜃𝑖 𝑐𝛼𝑖−1 𝑐𝜃𝑖 𝑐𝛼𝑖−1 −𝑠𝛼𝑖−1 −𝑑𝑖 . 𝑠𝛼𝑖−1
𝑖𝑇 = � � Eq. 2-3
𝑠𝜃𝑖 𝑠𝛼𝑖−1 𝑐𝜃𝑖 𝑠𝛼𝑖−1 𝑐𝛼𝑖−1 𝑑𝑖 𝑐𝛼𝑖−1
0 0 0 1

Using the data from Table 2-4 with Eq. 2-3 the following transformation matrices

can be found:

39
𝑐𝝋 𝑠𝝋 0 0
0 −𝑠𝝋 𝑐𝝋 0 0
1𝑇 = � � Eq. 2-4
0 0 1 𝐻
0 0 0 1

1 0 0 0
1 0 0 −1 𝑹
2𝑇 = � � Eq. 2-5
0 1 0 0
0 0 0 1

𝑐𝜽𝟏 −𝑠𝜽𝟏 0 0
⎡ ⎤
2 ⎢ −𝑠𝜽𝟏 /√2 −𝑐𝜽𝟏 /√2 1/√2 −𝐻 ⎥
3𝑇 = Eq. 2-6
⎢−𝑠𝜽𝟏 /√2 −𝑐𝜽𝟏 /√2 −1/√2 𝐻 ⎥
⎣ 0 0 0 1 ⎦
𝑐𝜽𝟐 −𝑠𝜽𝟐 0 0
3 𝑠𝜽𝟐 𝑐22.5 𝑐𝜽𝟐 𝑐22.5 −𝑠22.5 0
4𝑇 = � �
𝑠𝜽𝟐 𝑠22.5 𝑐𝜽𝟐 𝑠22.5 𝑐22.5 0 Eq. 2-7
0 0 0 1

1 0 0 0
4 0 𝑐22.5 −𝑠22.5 −𝒁𝑠22.5
5𝑇 = � � Eq. 2-8
0 𝑠22.5 𝑐22.5 𝒁𝑐22.5
0 0 0 1

𝑐𝝍 −𝑠𝝍 0 0
5 𝑠𝝍 𝑐𝝍 0 0
6𝑇 = � � Eq. 2-9
0 0 1 0
0 0 0 1
To find the end effector coordination in origin frame all these transformation

matrices must be multiplied:

𝒓𝟏𝟏 𝒓𝟏𝟐 𝒓𝟏𝟑 𝒑𝒙


0 0 1 2 3 4 5 𝒓𝟐𝟏 𝒓𝟐𝟐 𝒓𝟐𝟑 𝒑𝒚
6𝑇 = 1𝑇 2𝑇 3𝑇 4𝑇 5𝑇 6𝑇 = �𝒓 𝒓𝟑𝟐 𝒓𝟑𝟑 𝒑𝒛
� Eq. 2-10
𝟑𝟏
0 0 0 1

40
Elements of Eq. 2-10 are quite lengthy and tedious to work with. A MATLAB code

is presented in “Appendix A" to numerically and parametrically calculate the final

matrix in terms of φ, r, θ1, θ2, ψ and z.

For example, for following random configuration the result of forward

kinematics equation is:

𝜑 = 𝜃1 = 𝜃2 = 𝜓 = 0.5 𝑟𝑎𝑑𝑖𝑎𝑛𝑠

𝑅=𝑍=1

0.5992 −0.6263 0.4987 0.9781


𝟎 0.7534 0.6518 −0.0868 0.7908
𝟔𝑻 =� �
−0.2707 0.4277 0.8624 0.8624
0 0 0 1.0000

2.5 Inverse Kinematics

Although a completely general manipulator with 6 degrees of freedom does not

have a closed-form solution, certain special cases can be solved. For instance in this

design all axes have been chosen in a way that most of the parameters are zero.

Besides, a closed-form solution is always reachable when 3 consecutive axes

intersect at one point, based on Pieper’s work on 6 axis robots (Pieper and Roth

1969) (D. Pieper 1968). However, this work was mostly done on all revolute joints.

In this case a custom method is needed to solve the inverse kinematics since this

manipulator has two prismatic joints.

To approach an inverse kinematics problem two methods can be applied;

algebraic and geometric method. In algebraic approach kinematic equations must be

put into a form for which a solution is known. In geometric approach the spatial

41
geometry of the robot must be decomposed into several sub-problems to find the

solution for each degree of freedom one by one.

Forward kinematic equations of the proposed robot contain 12 kinematics

equations each of which is quite lengthy. For example, the first element 𝒓𝟏𝟏 in the

forward kinematic equations, is of the form:

𝒓𝟏𝟏 𝑟12 𝑟13 𝑝𝑥


0 𝑟21 𝑟22 𝑟23 𝑝𝑦
6𝑇 = � 𝑟 𝑟32 𝑟33 𝑝𝑧

31
0 0 0 1

𝒓𝟏𝟏 = sin 𝜑 ��sin 𝜃2 (. 271 cos 𝜃1 + .653) + .383 sin 𝜃1 (. 271 cos 𝜃1 − .653)(cos 𝜃2 − 1)�(cos 𝜓 cos 𝜃1

− .707 sin 𝜓 sin 𝜃1 )

+ �. 383 sin 𝜃1 sin 𝜃2

cos 𝜃1 − 1
− (. 271 cos 𝜃1 + .653)(. 271 cos 𝜃1 − .653)(cos 𝜃2 − 1)� �sin 𝜓 � �
2

+ .707 cos 𝜓 sin 𝜃1 �

cos 𝜃1 + 1
+ �sin 𝜓 � � + .707 cos 𝜓 sin 𝜃1 � �(. 271 cos 𝜃1 − .653)2
2

− cos 𝜃2 ((. 271 cos 𝜃1 − .653)2 − 1)� �

− cos 𝜑 ��sin 𝜃2 (. 271 cos 𝜃1 + .653)

cos 𝜃1 + 1
− .383 sin 𝜃1 (. 271 cos 𝜃1 − .653)(cos 𝜃2 − 1)� �sin 𝜓 � � + .707 cos 𝜓 sin 𝜃1 �
2

cos 𝜃1 − 1
− �sin 𝜓 � � + .707 cos 𝜓 sin 𝜃1 � (sin 𝜃2 (. 271 cos 𝜃1 − .653)
2

+ .383 sin 𝜃1 (. 271 cos 𝜃1 + .653)(cos 𝜃2 − 1) + (cos 𝜓 cos 𝜃1 − .707 sin 𝜓 sin 𝜃1 )

− cos 𝜃2 (. 146 sin2 𝜃1 − 1) − .146 sin2 𝜃1 ) �

There are 11 more equations like this to try to solve for the unknown angles and

prismatic variables. It is pretty obvious that there is no way to solve the inverse

42
kinematics of this manipulator using these types of trigonometric equations.

However, we can take advantage of the manipulator’s geometry and break the

inverse kinematics into a few sub-problems to find each joint parameter. This can be

done using the key points and axes described in the forward kinematics section.

In The approach first the key point of the mechanism which is the RCM needs to

be found. Then by having the RCM coordinate which can be directly linked to other

parameters of the design, the problem simplifies to a planar geometric problem to

solve for φ and R. In next step, Z which is the distance from the tool’s frame origin to

the RCM point can be calculated. The only unknowns at this step would be 𝜃1 , 𝜃2 and

𝜓. These parameters can be found using the concept of double-rotor mechanism. A

MATLAB code is available in the “Appendix B” that calculates the inverse kinematics

of this robot.

2.6 Workspace

The required workspace of this ophthalmic surgery robot is a spherical space

with 30 mm in diameter which is the size of the eyeball that must be reachable by its

end-effector. The reachable workspace of the robot is the 3D sub-space that the tip

of the robot can reach with at least one orientation. The shape of reachable

workspace for this robot is found by putting the upper limit and lower limit values

of each degree of freedom into forward kinematics equation which is resulted in a

shape illustrated in Figure 2-10.

Combination of the planetary ring motion and the top linear motion as the first

and second axes gives a workspace in shape of planar disc with its center located on

43
the surgical base plane and is coincident with the ring axis. This disc shapes the

locus of RCM.

The double-rotor mechanism (Axes III to VI) creates a conic workspace with two

portions based on the position of instrument tip. When tip of instrument is above

RCM which means Z value is positive the workspace is the upper right conic and

when it is below RCM and Z value is negative it is the lower left cone. This is shown

in Figure 2-11. Therefore the total reachable workspace can be found by moving the

conic workspace over this disc, which gives us the mushroom shape in Figure 2-10.

Figure 2-10: Reachable workspace of the robot.

44
Figure 2-11: Double-rotor conic workspace; positive and negative sections.

2.7 Accuracy

To find the integrated precision of this robotic system the precision of each axis

must be used which are calculated in section 2.3. Now that the forward kinematics

of the robot is available, the total accuracy of the system can be discussed.

To calculate the precision at the instrument tip, an initial home position is

considered. Then each DOF is misplaced by the amount equal to the precision of

each axis in a non-compensating direction, meaning that the misplacement in one

axis does not make up the error of the misplacement in another axis. Using forward

kinematics equation the final position and orientation of the tool tip are compared:

45
Table 2-5: Misplacement variables

State Joint variables


Home Position φ = θ1 = θ2 = 0
ψ=r=Z=0
Misplaced φ = 0.001° θ1 = 0.008° θ2 = 0.037°
ψ = 0.17° r = 0.0025mm Z = 0.0095mm

1 0 0 0
0 1 0 0
𝐻𝑜𝑚𝑒 = � �
0 0 1 0
0 0 0 1

0.9999933 −0.0036449 0.00034586 0.00000333


0.0036449 0.99999336 0.00000009 0.00250000
𝑀𝑖𝑠𝑝𝑙𝑎𝑐𝑒𝑑 = � �
−0.000346 0.00000116 0.99999994 0.00949999
0 0 0 1

The total tip misplacement is then:

∆𝑅 = �0.000003332 + 0.00252 + 0.009499992 = 0.00982344 𝑚𝑚 = 9.82 𝜇𝑚

46
CHAPTER 3 : CONTROL THEORY OF THE SYSTEM

3.1 Introduction to Theory of Control

Consider a system that receives an input u(t) and in response produces an

output y(t). The response y(t) is determined by the nature of the input signal u(t)

and by the nature of the system.

Figure 3-1: A simple input/output system.

If g(t) is assumed to be the response of the system to a unit impulse at time t=0,

then the response of the system y(t) to any input u(t) can be visualized as the

response to a train of impulses that approximates the function (see Figure 3-2).

Figure 3-2: Unit impulse 𝛿(𝑡) and response to a set of individual impulses.

47
Any one of the individual response curves can be expressed as 𝑢(𝜏)𝑔(𝑡 − 𝜏),

where 𝜏 is the time that the impulse is applied (R.Leigh 2004). Then the response to

any input u(t) can be given by the convolution integral using superposition:
𝑡
𝑦(𝑡) = � 𝑔(𝑡 − 𝜏)𝑢(𝜏)𝑑𝜏 = (𝑔 ∗ 𝑢)(𝑡) Eq. 3-1
0

If U(s) and G(s) are assumed to be the Laplace transform of u(t) and g(t)

respectively, then by taking Laplace transform of both sides of this equation, we get:

Eq. 3-2
Y(s)=G(s)U(s)

This is how Laplace transform makes it much simpler to find the response to any

input impulse u(t).

3.2 Closed Loop Transfer Function

A closed loop system is shown in Figure 3-3 where C is the controller, G is the

plant, r is the controller input or desired value and y is the plant output and the

transfer functions of these elements are C(s), G(s), R(s) and Y(s), respectively. The

system is linear and time invariant which means the elements of transfer functions

do not depend on time.

Figure 3-3: Closed loop system.

48
The characteristics of this system are defined as:

r: Controller Input

C(s): Controller Transfer Function

G(s): System Response

U(s): Controller Output

Y(s): System Output

E: Error

Error is the difference between system output and controller input:

E(s)= R(s)-Y(s) Eq. 3-3

Then based on the Eq. 3-2 driven in previous section, the following relations can

be expressed:

U(s)= C(s) E(s)

Y(s)= G(s) U(s)

Then, by substituting error by its definition from Eq. 3-3, we get:

U(s)= C(s) [R(s)-Y(s)]

Y(s)= G(s) C(s) [R(s)-Y(s)]

Eventually, the final result would be:

Y(s)= H(s) R(s)

G(s)C(s)
Where: H(s) = Eq. 3-4
1 + G(s)C(s)

H(s) is the transfer function of this closed loop control system.

49
3.3 Dynamic Model of a DC Motor

Since the robotic surgery system discussed in this thesis is composed of 6 DC

motors as the main elements of its control system, it’s essential to develop a linear

model to analyze the dynamic behavior of the system.

A DC motor as shown in Figure 3-4 consists of stator magnets that provide a

constant magnetic field, rotor coils that provide electromagnetic force for rotation.

The coils are powered from the commutator and the brushes. The more the

numbers of coils are the smoother the rotation gets. The geometrical characteristics

and position of the brushes (and the commutator of course) will be responsible for

changing the magnetic field of the electromagnets according to the position of the

rotor. The brushes are two metallic pieces that act like springs. On one side, they

have a piece of conductive material, usually made of carbon to stand against friction.

On the other side, they have the pin that the power supply is applied to the motor.

The brushes are pushed (by the spring action of the metallic part) against the

commutator. The commutator is a metallic ring, also conductive and is fixed on the

shaft of the motor. The shaft transfers the rotation to the rotating elements of the

mechanism.

Figure 3-4: A DC motor composition.

50
Also a DC motor can be modeled as an electric circuit of the armature with

mechanical load as shown in Figure 3-5 (Babuska and Stramigioli 1999).

Figure 3-5: A DC motor modeled as an electric circuit of the armature with

mechanical load.

The rotor and the shaft are assumed to be rigid. The parameters of this model

are defined as follows:

v: applied voltage

i: armature current

vb : EMF voltage

θ: angular position

ω: angular velocity

J: moment of inertia of the rotor

B: damping (friction) of the mechanical system

The input is the armature voltage v in Volts (driven by a voltage source). Measured

variables are the angular velocity of the shaft ω in radians per second, and the shaft

angle θ in radians.

51
For the armature electric circuit the following equation can be written based on

Kirchhoff law:

𝑑𝑖(𝑡)
𝑣(𝑡) = 𝑅𝑖(𝑡) + 𝐿 + 𝑣𝑏 (𝑡) Eq. 3-5
𝑑𝑡

Where, 𝑣𝑏 is back-electromotive voltage which is a voltage induced in the coil

due to its motion inside the magnetic field, as stated by Faraday’s law, R is the

armature resistance and L is its electric inductance.

The motor torque, T, is related to the armature current, i, by a constant factor𝑘𝑇 :

𝑇(𝑡) = 𝑘𝑇 𝑖(𝑡) Eq. 3-6

The back electromotive force (emf), 𝑣𝑏 , is related to the angular velocity by:

𝑑𝜃
𝑣𝑏 (𝑡) = 𝑘𝑏 𝜔(𝑡) = 𝑘𝑏 Eq. 3-7
𝑑𝑡

Where 𝑘𝑏 is the back-EMF constant. Also, for the power in this system the

equation Vi=𝑘𝑇 ω can be written, which results to 𝑘𝑇 = 𝑘𝑏 . Therefore, both these

constant factors can be substituted with simple K.

Also, the following equation can be written based on the Newton’s law:

𝑇(𝑡) = 𝐽𝜃̈(𝑡) + 𝐵𝜃̇(𝑡) + 𝑇𝑙 (𝑡) Eq. 3-8

While: 𝜔(𝑡) = 𝜃̇(𝑡)

Where, 𝑇𝑙 is the load torque.

Now that the equations of the system are derived, it’s time to take Laplace

transform of these equations. Using the Laplace transform, Eq. 3-5 and Eq. 3-7 can

be written as:

𝑉(𝑠) = 𝑅𝐼(𝑠) + 𝐿𝑠𝐼(𝑠) + 𝑉𝑏 (𝑠) Eq. 3-9

52
and 𝑉𝑏 (𝑠) = 𝐾𝛺(𝑠) Eq. 3-10

Where, s denotes the Laplace operator. Also, for Eq. 3-8 we get:

𝑇(𝑠) = 𝐽𝑠 2 𝜃(𝑠) + 𝐵𝑠𝜃(𝑠) + 𝑇𝑙 (𝑠) Eq. 3-11

From combination of Eq. 3-9 and Eq. 3-10, 𝐼(𝑠) can be expressed as:

𝑉(𝑠) − 𝐾𝑠𝜃(𝑠)
𝐼(𝑠) = Eq. 3-12
𝐿𝑠 + 𝑅

Since T(s) is also equal to K I(s), combining Eq. 3-11 and substituting I(s) from

Eq. 3-12, we get:

𝑉(𝑠) − 𝐾𝑠𝜃(𝑠)
𝐽𝑠 2 𝜃(𝑠) + 𝐵𝑠𝜃(𝑠) + 𝑇𝑙 (𝑠) = 𝐾 Eq. 3-13
𝐿𝑠 + 𝑅

From this equation, the transferred functions of shaft angle 𝜃(𝑠) and angular

velocity 𝛺(𝑠) would be found as follows:

𝑇(𝑠) − 𝑇𝑙 (𝑠)
𝜃(𝑠) = Eq. 3-14
𝑠(𝐽𝑠 + 𝐵)

𝑇(𝑠) − 𝑇𝑙 (𝑠)
𝛺(𝑠) = Eq. 3-15
𝐽𝑠 + 𝐵

53
Assuming the free load case (Tl (s)=0), Eq. 3-13 for the DC motor is shown in the

block diagram in Figure 3-6.

Figure 3-6: A block diagram of the DC motor.

The resulted transfer function for the input voltage, V(s), to the output angle,

θ(s), directly follows:

𝜃(𝑠) 𝐾
𝐺𝑎 (𝑠) = = Eq. 3-16
𝑉(𝑠) 𝑠[(𝐿𝑠 + 𝑅)(𝐽𝑠 + 𝑏) + 𝐾 2 ]

From the block diagram in Figure 3-6, it is easy to see that the transfer function

from the input voltage, V(s) , to the angular velocity, ω(s), is:

𝛺(𝑠) 𝐾
𝐺𝑣 (𝑠) = = Eq. 3-17
𝑉(𝑠) (𝐿𝑠 + 𝑅)(𝐽𝑠 + 𝑏) + 𝐾 2

Now that the position and velocity transfer functions, 𝐺𝑎 (𝑠) and 𝐺𝑣 (𝑠) are

derived, they can be used to analyze the behavior of this control system with real

motor parameters and specifications.

54
3.4 Motor (V) Transfer Function Analysis

The DC brushed motor (V) of the robot is chosen for further analysis. The

technical specifications of this motor are:

J= 0.13*10-7 (Kgm2)

R=27.4 (Ohm)

b=10-7 (Nms/rad)

L=0.399*10-3 (H)

K=0.01 (Nm/A)

The position transfer function for this motor would be:

0.01
𝐺𝑎 (𝑠) =
5.187 ∗ 10−12 𝑠 3 + 3.562 ∗ 10−7 𝑠 2 + 0.1027 ∗ 10−3 𝑠

Also, the velocity transfer function is:

0.01
𝐺𝑣 (𝑠) =
5.187 ∗ 10−12 𝑠 2 + 3.562 ∗ 10−7 𝑠 + 0.1027 ∗ 10−3

The above transfer functions can be entered into the Matlab code which is

presented in “Appendix C” to plot the responses for the velocity and angle position.

This code takes the position and velocity transfer functions and returns the plots of

response of the system to step and impulse inputs.

55
3.4.1 Step and Impulse Responses

The step and impulse responses of the system are the first characteristic of the

system to be analyzed. The step response is the response of the system to a unit

input at t>=0 which here means to connect the motor to a power source that

produces one volt. On the other hand, the impulse response is an infinite input at

time t=0 and zero otherwise. Note that the Laplace transform of the step function is

equal to one over s, [V(s)step=1/s], while the Laplace transform of the impulse

function is one, [V(s)impulse=1]. Also, by definition, the derivation of the step function

is the impulse function.

The step and impulse responses of the position transfer function are shown in

Figure 3-7.

Step Response
From: Voltage To: Angle
3
Position (rad)

0
0 0.005 0.01 0.015 0.02 0.025 0.03
Time (seconds)
Impulse Response
From: Voltage To: Angle
100
Position (rad)

50

0
0 0.005 0.01 0.015 0.02 0.025 0.03
Time (seconds)

Figure 3-7: The step and impulse responses of the position transfer function.

After giving the input of unit voltage to the motor the step response plot is rising

as it is expected due to increase in angular position when the motor starts rotating.

56
However, the response of the position to the impulse input is converging to 100 rad,

which can be validated by taking the inverse Laplace transfer of 𝐺𝑎 (𝑠) when V(s)=1,

in order to find 𝐺𝑎 (𝑡) and plot the result.

Also, same responses are plotted for velocity transfer function which is shown in

Figure 3-8.

Step Response
From: Voltage To: Velocity
100
Velocity (rad/s)

50

0
0 0.005 0.01 0.015 0.02 0.025 0.03
Time (seconds)
Impulse Response
4
x 10 From: Voltage To: Velocity
3
Velocity (rad/s)

0
0 0.005 0.01 0.015 0.02 0.025 0.03
Time (seconds)

Figure 3-8: The step and impulse responses of the velocity transfer function.

The step response of velocity shows the same behavior as the impulse response

of angular position which is expected since the velocity is related to angular position

by Ω(s) = sθ(s) and the Laplace transfer of step function is related to the Laplace

transfer of the impulse function by V(s)step=s V(s)impulse.

𝛺(𝑠) sθ(s)
𝐺𝑣 (𝑠) = =
V(s)step s V(s)impulse

The impulse response of velocity also shows an expected behavior by a rapid

rise in velocity at instant of impulse and then going back to zero.

57
3.4.2 Frequency Domain Analysis

In order to analyze the frequency response of the system, the “Bode diagram” is

plotted as can be seen in Figure 3-9.

Bode Diagram
From: Voltage To: Angle
200
Magnitude (dB)

-200

-400
-90
Phase (deg)

-180

-270
1 2 3 4 5 6 7
10 10 10 10 10 10 10
Fre que ncy (rad/s )

Bode Diagram
From: Voltage To: Velocity
50
Magnitude (dB)

System: Gv
0 I/O: Voltage to Velocity
Frequency (rad/s): 43.8
-50 Magnitude (dB): 39.7

-100
0
Phase (deg)

-90

-180
1 2 3 4 5 6 7
10 10 10 10 10 10 10
Fre que ncy (rad/s )

Figure 3-9: The position and velocity responses of the system to the voltage

frequency.

A Bode Plot is a useful tool that shows the gain and phase response of a given

linear time invariant system for different frequencies. The frequency of the bode

plots are plotted against a logarithmic frequency axis. Every tick mark on the

frequency axis represents a power of 10 times the previous value. The bode

magnitude plot measures the system input/output ratio in special units

called decibels, which is in this case can be expressed as: 𝑑𝐵 = 20𝑙𝑜𝑔𝐺(𝑠).

58
The Bode phase plot measures the phase shift in output with respect to input in

degrees. The reason to observe the frequency domain of the system is to make sure

there would not be any critical frequency to create excitation in the system.

As it can be seen in these diagrams, the bode magnitude goes down as the

frequency increases for both angle and velocity responses. To evaluate the result,

for example for frequency between 10 and 100 the magnitude of the velocity

response is almost constant and approximately equal to 40 dB which agrees with

following calculations:

0.01
𝑑𝐵 = 20𝑙𝑜𝑔𝐺𝑣 (𝑠) = 20 𝑙𝑜𝑔 ≅ 20𝑙𝑜𝑔100 = 40 𝑑𝐵
5.187 ∗ 10−12 𝑠 2 + 3.562 ∗ 10−7 𝑠 + 0.1027 ∗ 10−3

Also, from these diagrams it is assured that there is no critical frequency and

excitation risk to threaten the system.

3.5 Proportional Controller

Extra space

Now that the behavior of the system has been observed through step, impulse

and frequency responses, it’s time to add a controller. First, a simple proportional

controller would be added to the system as shown in Figure 3-10.

Figure 3-10: A simple closed loop system with proportional controller.

59
In this system, 𝜃𝑑 (𝑠) is the desired angular position, Kp is the proportional gain

and 𝜃(𝑠) is the actual position. The error is defined as the difference of actual

position and desired position:

𝐸(𝑠) = 𝜃𝑑 (𝑠) − 𝜃(𝑠)

In the proportional controller the error term is multiplied by proportional gain,

Kp, to produce the controller output:

𝑈(𝑠) = 𝐾𝑝 ∗ 𝐸(𝑠)

Therefore, the closed loop transfer function of proportional controller system

according to Eq. 3-4 is:

𝐾𝑝. 𝐺(𝑠)
𝐻(𝑠) = Eq. 3-18
1 + 𝐾𝑝. 𝐺(𝑠)

Then, for the motor 5 that is being discussed in this chapter, the position transfer

function of proportional controller will be as followed:

0.01𝐾𝑝
𝐻𝑎 (𝑠) = Eq. 3-19
5.187 ∗ 10−12 𝑠 3 + 3.562 ∗ 10−7 𝑠 2 + 0.0001027𝑠 + 0.01𝐾𝑝

If a big value is been chosen for proportional gain it will result in a large change

in the output for a given change in the error. If the proportional gain is too high, the

system can become unstable. In contrast, a small gain results in a small output

response to a large input error, and a less responsive or less sensitive controller. If

the proportional gain is too low, the control action may be too small when

60
responding to system disturbances. The goal is to find the best value for Kp that

minimizes the error and the response time. To find the best Kp, the step response of

this system for Kp=1, 5, 10 and 50 has been observed.

Step Response
From: Desired Angle To: Angle
1.4

1.2

1
Position (rad)

0.8

0.6

0.4

0.2

0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
Time (seconds)

Figure 3-11: The proportional controller step response for Kp=1.

For the first plot the value of Kp assumed to be equal one, Kp=1, the result shows

a quite small overshoot while the system very slowly approaches to the desired

value. So, the proportional gain must be increased to give a faster approach.

The next response is plotted for Kp=5 as shown in Figure 3-12. Here, the result

shows a faster approach to the desired response while gives a bigger over shoot at

first which is about 1.27 times of the desired value.

61
Step Response
From: Desired Angle To: Angle
1.4

1.2

Position (rad)
0.8

0.6

0.4

0.2

0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
Time (seconds)

Figure 3-12: The proportional controller step response for Kp=5.

To further the study of the effect of proportional gain, the results for Kp=10 and

50 are being plotted in Figure 3-13 and Figure 3-14.

Step Response
From: Desired Angle To: Angle
1.5

1
Position (rad)

0.5

0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
Time (seconds)

Figure 3-13: The proportional controller step response for Kp=10.

62
Step Response
From: Desired Angle To: Angle
1.8

1.6

1.4

1.2

Position (rad)
1

0.8

0.6

0.4

0.2

0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
Time (seconds)

Figure 3-14: The proportional controller step response for Kp=50.

As it can be seen in these diagrams, by increasing the proportional gain larger

overshoots will be resulted while the response time goes down.

The proportional controller does not deliver a proper performance for the high

precision robot which is the interest of this search. Therefore a more advanced

control system must be considered.

3.6 PID Controller

PID is an acronym for proportional- integral- derivative control system. This is

the most common controller used in many industries. The structure of this three

terms feedback controller helps to attain a set point irrespective of disturbances or

any variation in characteristics of the plant of any form. It calculates its output based

on the measured error and the three controller gains; proportional gain Kp, integral

gain Ki, and derivative gain Kd. The proportional gain simply multiplies the error by

63
a factor Kp. This reacts based on how big the present error is. The integral term is a

multiplication of the integral gain and the sum of the recent errors. The integral

term helps in getting rid of the accumulation of past errors to eliminate the residual

steady state error and causes the system to catch up with the desired set point. The

derivative controller determines the reaction to the rate of which the error has been

changing. It helps with prediction of future errors based on current rate of change,

improves settling time and stability of the system.

The final output of the PID controller (U) is calculated using the following

equation:

𝑑
𝑢(𝑡) = 𝐾𝑝 ∗ 𝑒(𝑡) + 𝐾𝐼 � 𝑒(𝑡)𝑑𝑡 + 𝐾𝐷 𝑒(𝑡) Eq. 3-20
𝑑𝑡

Figure 3-15: The block diagram of closed loop PID controller.

64
3.6.1 Transfer Function of PID Controller

The transfer function of a PID controller is found by taking the Laplace

transform of Eq. 3-20:

𝐾𝐼
𝑈(𝑠) = 𝐾𝑝 𝑒(𝑠) + 𝑒(𝑠) + 𝐾𝐷 𝑠𝑒(𝑠)
𝑠

Also can be expressed as:

𝐾𝐼
𝑈(𝑠) = �𝐾𝑝 + + 𝑠 𝐾𝐷 � 𝑒(𝑠) = 𝐶(𝑠)𝑒(𝑠)
𝑠

Where:

𝐾𝐼
𝐶(𝑠) = 𝐾𝑝 + + 𝑠 𝐾𝐷
𝑠

C(s) is called the PID controller transfer function. Combining this result with

Eq. 3-4, the position transfer function of PID controller is found as follows:

𝜃𝑑 (𝑠) 𝐶(𝑠). 𝐺(𝑠)


𝐻(𝑠) = = Eq. 3-21
𝜃(𝑠) 1 + 𝐶(𝑠). 𝐺(𝑠)

With this closed loop transfer function the behavior of the system can be

analyzed in order to find the best proportional, integral and derivative gains. This

process is called manual tuning. There are a couple of strategies on how PID can be

tuned; this includes trial and error tuning method, Ziegler-Nichols tuning method

(Meshram and Kanojiya 2012) etc. The goal is to minimize settling time with no over

shoot. In this case the best optimum result is obtained by trial and error.

65
3.6.2 Proportional Gain Effect

To analyze the effect of proportional gain in PID system, the integral and

derivative gains assumed to be Ki=100 and Kd=1. The results are plotted for Kp=1,50

and 500.

Figure 3-16: The PID controller step response for Kp=1 (left) and Kp=50 (right).

PID Step Response


From: Desired Angle To: Angle
1.4

1.2

1
Position (rad)

0.8

0.6

0.4

0.2

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
-3
Time (seconds) x 10

Figure 3-17: The PID controller step response for Kp=500.

These plots show even though the system is sensitive to the value of Kp, but the

result does not significantly change when Kp value is changed from 1 to 50 and even

66
to 500. As it was discussed earlier, increasing Kp results in shorter settling time and

larger overshoot.

3.6.3 Integral Gain Effect

To study the effect of integral gain in PID system, the proportional and derivative

gains assumed to be Kp=100 and Kd=1. The results are plotted for Ki=1 and 100.

PID Step Response


From: Desired Angle To: Angle
1.4

1.2

1
Position (rad)

0.8

0.6

0.4

0.2

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
-3
Time (seconds) x 10

Figure 3-18: The PID controller step response for Ki=1.

PID Step Response


From: Desired Angle To: Angle
1.4

1.2

1
Position (rad)

0.8

0.6

0.4

0.2

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
-3
Time (seconds) x 10

Figure 3-19: The PID controller step response for Ki=100.

67
In this case, even with a big change in value of Ki from 1 to 100, it does not show

any significant impact on quality of response.

3.6.4 Derivative Gain Effect

To study the effect of derivative gain in PID system, the proportional and integral

gains assumed to be Kp=100 and Ki=10. The results are plotted for Kd=0, 1 and 2.

Figure 3-20: The PID controller step response for Kd=0 (left) and Kd=1 (right).

PID Step Response


From: Desired Angle To: Angle
1.4

1.2

1
Position (rad)

0.8

0.6

0.4

0.2

0
0 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.009 0.01
Time (seconds)

Figure 3-21: The PID controller step response for Kd=2

As it can be seen in these diagrams, the response of the system is highly sensitive

to the value of Kd that even a very small changes in this value have a big impact on

quality of response.

68
3.6.5 PID Controller Gain Values

The best response for this system is obtained with Kp=200, Ki=100 and Kd=0.7.

As it can be seen in Figure the response reached with almost no over shoot (0.01%)

in a very short time.

PID Step Response


From: Desired Angle To: Angle
1.4

1.2

1
Position (rad)

0.8

0.6

0.4

0.2

0
0 0.5 1 1.5 2 2.5 3 3.5 4
-3
Time (seconds) x 10

Figure 3-22: The best PID response.

These are the best PID gains could be found for this specific system. If Kp is

chosen to be smaller than this value, the system gets to the set point slower and if a

bigger value is selected for Kp the over shoot would be resulted. However the system

is not pretty much sensitive to the value of Ki, still this value of Ki gives the faster

result. If the value of Kd is chosen to be slightly less than this value, like 0.6, it will

result in steady state error, and if it is chosen to be slightly more than this value, like

0.8, it will result in over shoot.

69
CHAPTER 4 : CONTROLLER ARCHITECTURE

In this chapter the control system of the robot with its all elements and features

will be discussed. The control system is composed of the encoder, controller

module, Controller Area Network (CAN), and controller software, as shown in

Figure 4-1.

Figure 4-1: The microsurgery robotic system with its motors and controllers

connected through the USB port to the controller software.

4.1 Encoder

The incremental encoder is used for position and speed evaluation (Kafader

2014). Incremental digital encoder is one of the most common sensor type for

70
measuring speed and position in micro-drive tools. Many various designs with

different working principles are using this kind of encoder, but all of them basically

subdivide one revolution into a number of steps (increments), and send signal

pulses to the controller each time an increment is detected. Incremental encoders

generally supply square-wave signals in two channels, A and B, which are offset by

one quarter signal length (or 90 electrical degrees).

In this robot the encoder pulse for motors (I) to (V) is 512 cpt (counts per turn)

and 128 cpt for motor (VI). The high resolution provided by these encoders gives

the robot the ability of the micro positioning control. The encoder part and its

connection cable for motor (I) are shown in Figure 4-2 by red arrows.

Figure 4-2: Motor (I) encoder and its cable

The characteristic parameter of the encoder is the number of square pulses N

per revolution, given in cpt (counts per turn). By counting the number of state

transitions in both channels, the real resolution is four times higher. These states

71
are called quadrature counts or quad counts (qc) and are used as the position unit in

EPOS (the controller software) systems.

The higher the resolution, the more accurately the position can be measured,

and the more accurately the speed can be derived from the change in position as a

function of time. The direction of rotation follows from the sequence of the signal

pulses of channel A and B: Channel A leads in one direction, channel B leads in the

other direction. As shown in Figure 4-3.

Figure 4-3: The signals of an incremental encoder.

Position evaluation is determined by counting the signal edges (increments)

from a defined reference position e.g. home position. The unit of positioning is quad

counts (qc).

Speed evaluation is calculated from the position change per sampling interval. In

this case, the natural unit of speed is quad counts per millisecond. Therefore, speed

can only be expressed in steps of 1qc/ms, which is usually translated in the more

72
common and practical unit rpm. Acceleration and deceleration values are given in

the practical unit rpm/s.

A high encoder pulse count N not only results in a high position resolution, but

also improves the control dynamics. Feedback information about a change in the

actual position value is obtained more quickly and the control circuit can initiate

corrective action much faster.

4.2 Controller Module

All six controllers in this robotic system are modular constructed digital

positioning controller. It is suitable for DC and EC motors with incremental encoder

with a power range from 1 to 700 watts. A number of operating modes provides

flexible application in a wide range of drive systems in automation technology and

mechatronics. Motion control is mechatronics; it is the combination of mechanics

and electronics of actuators (motors) and sensors all controlled by a software.

The controller device used in this robotic control system is a CANopen motion

controller device. CANopen is a standard maintained and supervised by the

independent user organization CAN in Automation (CiA). A Controller Area

Network, or CAN for short, is a common message based protocol used in control

systems (CAN in Automation (CiA) 1992). It is also used in industrial/robotic

applications such as motor control.

The following discussion is to provide a better understanding of how

communication is organized in this system.

73
On a physical level, each device - called a node and given its distinctive address

(node number) - is mounted in parallel on the bus (Kohanbash 2014). While there is

no standard way of wiring a CAN bus there are always two wires/signals, and often

another two. The first two are CAN+ (or CAN_H) and CAN- (or CAN_L) which are

used for sending the CAN data. The next two wires are for power and ground that

are connected to devices on the network. All of CAN devices can be wired in series

creating a large bus of devices. The bus network needs proper termination on both

sides to avoid signal reflections, which means having a 120 Ohm resistance at each

end. A physical layout of this robot with 6 nodes is shown in Figure 4-4.

Figure 4-4: Physical layout of the six motor controllers CAN bus.

Signal bits require a certain time to spread over the whole bus. Therefore, the

transmission rate depends on the bus length. Typically in robotics each CAN

network will have one “master” node which is typically a computer or micro-

controller that requests information or listens for alerts/emergencies from all of the

CAN devices on the network. In this robot master node is node 1 which is the

74
controller of motor I. This node sends and receives signals from other 5 controllers

and communicates that to the software (computer).

There are two sets of controller used in this design. The first one is “EPOS2 24/2

for Maxon EC motors” which is used for first four motors and the other one is

“EPOS2 24/2 for Maxon DC motors” which is used for motor (V) and (VI).

4.2.1 Maxon Controller for EC Brushless Motors

The EC motor controller with its connection ports is shown in Figure 4-5. This is

the controller which is used to control first four motors of the robot.

Figure 4-5: The EC brushless controller, interfaces and designations

J1 connectors are for supply and control signals which its pins 12 and 13 are

wired to the power supply. J2 connectors are communication connector that is used

for CAN bus communication (Maxon, Application Notes Collection 2013). This

connectors are used for send/receive control signals. Pin 1, 2 and 5 are wired to

CAN high bus line, CAN low bus line and ground line, respectively. J8 is motor/hall

75
sensors connector. J9 is for the encoder cable connection and finally J15 is the USB

port to connect the controller to the computer.

JP1 is CAN node identification and is used to set CAN ID or node address (Maxon,

Application Notes Collection 2013). The CAN ID is set with DIP switches 1 to 4 as

shown in Figure 4-6. With these four binary switches, 15 addresses (1 to 15) may be

coded.

Figure 4-6: JP1 – Numbering scheme

The CAN ID results in the summed values of “ON” DIP switch addresses. For

instance DIP switch setting to identify the six nodes of this robotic system must be

like the way shown in the Table 4-1. DIP switches 5 and 6 do not have any impact on

the CAN ID.

76
Table 4-1: DIP switch setting for robotic system with 6 nodes.

CAN ID DIP Setting Calculation

1 20

2 21

3 20+21

4 22

5 20+22

6 21+22

DIP switch 5 is to detect CAN bit rate, and DIP switch 6 is CAN bus termination.

The CAN bus must be terminated at both ends by a termination resistor of 120 Ω,

typically. Depending on utilization of the controller, individual CAN bus termination

settings must be performed. Using DIP switch 6, the controller-internal bus

termination resistor can be activated/deactivated.

Also there is a Status LED on the controller board to display the current status of

the controller as well as possible errors. Green LED shows the operating status

77
while Red one indicates errors. Figure 4-7 shown the controller (III) with the Green

LED “ON”.

Figure 4-7: Controller (III) in operating status with the Green LED “ON”.

Maxon Controller for DC Brushed Motors

The DC motor controller with its connections is shown in Figure 4-8. This is the

controller used for motors (V) and (VI) of the robot.

Figure 4-8: DC brushed controller, interfaces and designations.

78
The J1, J2, J15 and JP1 connectors are the same as discussed in the previous

section. J3 is the motor/encoder connector which connects to the motor and

encoder through Adapter encoder connector as shown in the Figure 4-9 with motors

(V) & (VI) controller wired to the system.

Figure 4-9: Motor/encoder connector (left) and DC brushed controllers (V) & (VI)

wired to the system (right).

4.3 Controller Area Network (CAN)

The existing robot control system structures can be roughly divided into two

categories, centralized control and decentralized control (Kazanzides 2009). As it

was already mentioned, this robotic system benefits from CAN as its control system

which is counted as a distributed or decentralized control system where each

controller installed on the related joint of the robot and connected to the master

processor devices through dedicated wiring. Clearly decentralized control provides

more flexibility, asynchronous operation and parallel processing.

79
By placing the controllers on the robot arm, significant reductions in cabling

could be achieved. In particular, the thick cables containing multiple wires for motor

power and sensor feedback could be replaced by thin network and power cables.

The other reason that makes CAN network the best option for this case is that it

provides an inexpensive, durable network that helps multiple CAN devices

communicate with one another. The advantage of this is that electronic control units

(ECUs) can have a single CAN interface rather than analog and digital inputs to every

device in the system which decreases overall cost and weight of the system.

The other advantage of CAN is its advance Broadcast Communication ability.

Each of the devices on the network has a CAN controller and is therefore intelligent.

All devices on the network see all transmitted messages. Each device can decide if a

message is relevant or if it should be filtered. This structure allows modifications to

CAN networks with minimal impact. Additional nodes can be added without

modification to the network.

Also, CAN system is very well capable of handling errors. Frames with errors are

disregarded by all nodes, and an error frame can be transmitted to signal the error

to the network. Global and local errors are differentiated by the controller, and if

too many errors are detected, individual nodes can stop transmitting errors or

disconnect itself from the network completely.

80
4.4 Easy Positioning System (EPOS) Controller Software

The EPOS Studio is the graphical user interface for all our Maxon controllers

(Kafader 2014). It is capable of communicating with a motion controller with a

built-in programmable logic controller (PLC).

Essentially, the controller contains two devices, a PLC and a motion controller,

that are internally linked by a CAN bus. The EPOS Studio on the computer

communicates with these two devices by means of a USB2.0 serial connection as the

default communication. The PLC part is the master of the device. It controls the

process flow and all slave units in the network; one of the slave units being the built-

in motion controller as shown in Figure 4-10.

Figure 4-10: Schematic overview of the controller with external connections and

communication.

To get connected to the controllers through the EPOS Studio on the computer,

first we need to wire all 6 controllers together as it was discussed in section 4.2.

81
Then to connect this series of controller to the software, one of them need to be

chosen (usually the first motor controller) to be connected through USB port to the

computer. This one is going to be the master controller.

To get started, we need to create a new project. The controller hardware of the

project is displayed in the Workspace tab of the Navigation.

Figure 4-11: The EPOS Studio screen with the Workspace tab open.

The Communication tab shows the setup of the project from a communication

point of view. The nice thing about the EPOS Studio is that we can directly

communicate with any device in the network once we get connected to them. The

first node connected to the computer through USB, usually is quickly recognized by

the software. To find the rest of nodes on the network, there is a scanning device

available on communication tab which scans for up to 127 nodes, but once it found

82
the rest of our 5 nodes, we can quit the scanning project and add those nodes to the

network.

Figure 4-12: Communication set up with six nodes.

4.4.1 The Motion Controller

Now that we have configured the system the EPOS motion controller is now

ready for further exploration. The tools for the exploration of the motion controller

can be found in the Tools tab of the EPOS Studio.

The standard positioning mode for most applications is the “Profile Position

Mode”. To execute a point-to-point movement, we need to use this option and enter

required motion parameters such as target position, velocity and acceleration.

As it was discussed in section 4.1, position evaluation is determined by counting

the increments from a defined reference position. The unit of positioning is quad

counts (qc) which equals four encoder pulse.

83
Figure 4-13: Different operating modes under the “Tools tab”.

The required calculation in order to convert the motion of six axes of the robot

into the qc unit (which is the acknowledgeable unit by the EPOS Studio program) is

summarized in Table 4-2.

For example to drive the robot arm around the ring for 45 degree, the target

position for node 1 (the first Axis- rotating axis) would be 116736 “qc” which is

resulted from:

4 ∗ 𝐸𝑛𝑐𝑜𝑑𝑒𝑟 𝑃𝑢𝑙𝑠𝑒 ∗ 𝐺𝑒𝑎𝑟 𝑏𝑜𝑥 𝑟𝑎𝑡𝑖𝑜 ∗ 45 (𝑡𝑎𝑟𝑔𝑒𝑡 𝑝𝑜𝑠𝑖𝑡𝑖𝑜𝑛)


15 (𝑡ℎ𝑒 𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒 𝑡𝑟𝑎𝑣𝑒𝑙𝑒𝑑 𝑒𝑎𝑐ℎ 𝑝𝑖𝑛𝑖𝑜𝑛 𝑡𝑢𝑟𝑛)

For the second axis which is a linear one, to move the arm 20 mm along its axis

the target position would be 16126 “qc” which is calculated from:

4 ∗ 𝐸𝑛𝑐𝑜𝑑𝑒𝑟 𝑃𝑢𝑙𝑠𝑒 ∗ 𝐺𝑒𝑎𝑟 𝑏𝑜𝑥 𝑟𝑎𝑡𝑖𝑜 ∗ 20 (𝑡𝑎𝑟𝑔𝑒𝑡 𝑝𝑜𝑠𝑖𝑡𝑖𝑜𝑛)


2 ∗ 1.27 (𝑡ℎ𝑒 𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒 𝑡𝑟𝑎𝑣𝑒𝑙𝑒𝑑 𝑝𝑒𝑟 𝑡𝑢𝑟𝑛)

84
Figure 4-14: Data entry for 45 degree motion of node 1 (around the ring) in profile

position mode.

85
Table 4-2: Data sheet to convert the motion of six axes to quad count (qc) unit.

Designed Encoder Input Data


Motor Gear box Traveling Distance (each turn) Quad Counts (qc)
Range Pulse Angle/Distance
Each pinion turn
Axis I 512 19 15 travel 15° around the 45 ° 4*512*19*45/15 116736
ring
-45° to
225°

Axis II 512 1 1.27 mm per turn 20 mm 4*512*20/2*1.27 16125.98


0 to 90
mm

Axis III 512 84 1 ° 90 ° 4*512*84*90/360 43008


-180° to

86
180°

Axis IV 512 19 1 ° 90 ° 4*512*19*90/360 9728


-180° to
180°

Axis V 512 1 1.22 mm per turn 10 mm 4*512*10/2*1.22 8393.443


-30 to 30
mm

Axis VI 128 16 1 ° 180 ° 4*128*16*180/360 4096


-180° to
180°
4.4.2 Position Accuracy

To move an object to a specified final position, the object must first be set in

motion (accelerated), moved and then decelerated. Finally, the specified position

must be maintained against any interfering forces. All these operations require a

motor torque that must be made available to the position control system. Ultimately,

position control relies on the lower level current control loop for creating the

necessary torque.

Here what happens is the master system sends motion commands to the motion

controller. They are processed by the path generator that calculates intermediate

positions at given time intervals (1ms on EPOS systems) on the path to reach the

final position. These set values are transmitted to the position control loop, which,

by comparison with the actual position, determines the set (or demand) values for

the current controller. In the EPOS the current control loop has a cycle time of 0.1

ms or 10 kHz.

On a well-tuned system, end positions are reached to within 1 encoder quad

count (qc). As soon as the position is more than 1 qc off the target position control

loop will take corrective actions.

It must be noted that, the system reaction on deviations from the required

positions depends on the torque and speed capabilities of the system and also on the

friction, mass inertias and the set feedback and feed-forward parameters. The

controller parameter values have been established during the tuning process based

on the system properties during tuning process.

87
The other important position accuracy aspect is how the target position is

reached. Again, the results depend on how the system was tuned and which aspect

of accuracy was given the most weight during the tuning: e.g. no position overshoot

permitted, or very fast motion regardless of overshoot. Based on the application of

the device in high precision microsurgery, obviously the tuning as it was discussed

in chapter three has been done with no overshoot permitted.

88
CHAPTER 5 : CONCLUSION

In this chapter, the main conclusions are briefly discussed and recommendations

for future work are presented.

5.1 Conclusions

In this thesis work, first it was explained how robotic surgery can benefit

ophthalmic surgery, and then different ophthalmic surgeries were studied in order

to understand required workspace, function and precision. Based on those

requirements a robotic system with 6 DOF is proposed and designed.

This system features a 350 mm ring that can hold up to three different instrument

manipulators and provide a rotary motion about the optic axis of the eye, φ, a linear

motion in the radial direction that determines the entry point to the eye, r, the

double-rotor mechanism that provides a remote center of motion for the surgical

manipulator at the entry point to the eye with two lateral rotations about the entry

point to the eye θ1 and θ2, a stroke motion along the instrument axis, Z, and a

rotation about the instrument axis, ψ.

This design is compact and offers wide range of motion and is capable of adjusting

remote center of motion. The theoretical accuracy of the system is calculated to be

10 μm. Also, forward and inverse kinematics of the system were defined.

89
Since the robotic surgery system discussed in this thesis is composed of 6 DC

motors as the main elements of its control system, a linear model to analyze the

dynamic behavior of the system was developed.

To run 6 DOF system, six miniature DC motors with PID controllers were used. To

perform the motion control of the system, first the dynamic behavior of a DC motor

was studied and modeled and then the position and velocity transfer functions were

derived. By substituting real motor parameters and specifications into those

transfer functions, the control diagrams were plotted and used to study the

behavior of the system. Then, a proportional controller was introduced to the

system and the process of finding the best proportional gain to give the optimum

response with fast convergence and no overshoot was investigated. To manually

tune the system, the same process was repeated to find the effect of integral and

derivative gains. Eventually, the PID control system with the optimum gains was

found and the result was presented.

The elements of the control system including encoder, controller modules,

Controller Area Network (CAN) and the controller software were discussed next. To

organize communications among these elements, a network which is here called

CAN bus needed to be created. In order to do so an ID number must be assigned to

each motor controller as a node on the system and then it must be properly

connected to the rest of nodes on network. The method of node number assignment

and wiring is also described.

90
The EPOS Studio is the graphical user interface that can provide connection with

controllers in order to run the motors. This program offers a wide range of control

options and provides a very well defined master for this robotic system.

Motion commands must be expressed in understandable unit which is here called

quadratic counts (qc). This unit is the smallest motion achievable by this design that

pretty much guarantees the required precision for intended surgeries. The

procedure to convert all sorts of motions to qc unit is tabulated and presented.

5.2 Recommendations for Future Work

The surgical robot which is the purpose of this thesis is fully assembled and

motorized and ready to function at this point. The master control system offers

options to adjust the magnitude of velocity, acceleration and deceleration in order to

produce smooth motion on all joints.

As mentioned earlier in the body of this thesis, this robot is the slave robot,

which does the operation. The input to the robot should be generated by a master

robot. The main part of the future work is to design a master robot and the

communication method between these two robots. Although designing such system

requires another comprehensive research, the studies on the eye surgery

procedures in this work can be used as a baseline for designing an efficient and

ergonomic master console.

Since the forward and inverse kinematics of the slave robot is fully determined,

the mapping between surgeon moves on the master console and slave robot

manipulation could be achieved with high accuracy.

91
The control system that receives the input from the master console and translate

it to slave robot movement can be designed in a way to scale the surgeon’s hand

motion down and also reduce the tremor.

The next step would be design verification and validation based on regulatory

agents standards such as FDA. The results of the verification and validation tests can

be used to optimize the design of the proposed slave robot.

92
BIBLIOGRAPHY

American Society of PeriAnesthesia Nurses. PeriAnesthesia Nursing Core Curriculum:


Preprocedure, Phase I and Phase II PACU Nursing. 2nd. Edited by Lois Schick
and Pamela E. Windel. Saunders Elsevier, 2012.

Ang, W. T., C. N. Riviere, and P. K. Khosla. "An active hand-held instrument for
enhanced microsurgical accuracy." MICCAI, 2000: 878 – 886.

Babuska, Robert, and Stefano Stramigioli. "Matlab and Simulink for Modeling and
Control." 1999.

Biographicsmedia. Eye anatomy drawing sketch image illustration. 2006.


http://www.biographixmedia.com/human/eye-anatomy.html (accessed
2013).

Bourla, D. H., J. Hubschman, M. Culjat, A. Tsirbas, A. Gupta, and S. D. Schwartz.


"Feasibility study of intraocular robotic surgery with the da vinci surgical
system." Retina 28, no. 1 (2008): 154-158.

Buckley, Sean. iRobot, InTouch Health unveil RP-VITA. 2012.


http://www.engadget.com/2012/07/24/irobot-intouch-health-unveil-rp-
vita-telepresence-robot/ (accessed 2013).

Butter, M., et al. Robotics for healthcare, Final report. TNO Informatie en
Communicatietechnologie: TNO, The Netherlands Vilans, The netherlands
EuroAct, Japan FhG ISI, Germany VTT, Finland, 2008.

CAN in Automation (CiA). March 1992. http://www.can-cia.org/.

Charles, S., et al. "Dexterity-enhanced telerobotic microsurgery." International


Conference on Advanced Robotics (ICAR), 1997: 5-10.

Craige, J. J. Introduction to Robotics: Mechanics and Control. 3rd. New Jersey: Pearson
Prentice Hall, 2004.

93
Dalvand, Mohsen Moradi, and Bijan Shirinzadeh. "Motion control analysis of a
parallel robot assisted minimally invasive surgery/microsurgery system
(PRAMiSS)." Robotics and Computer-Integrated Manufacturing 29, no. 2
(2013): 318-327.

Das, H., H. Zak, J. Johnson, J. Crouch, and D. Frambach. "Evaluation of a telerobotic


system to assist surgeons in microsurgery." Computer Aided Surgery 4, no. 1
(1999): 15-25.

Diana, Alison. 12 Advances in Medical Robotics. January 27, 2011.


http://www.informationweek.com/healthcare/patient-tools/12-advances-
in-medical-robotics/d/d-id/1095720? (accessed November 1, 2013).

Donelson, S. M., and C. C. Gordon. Matched anthropometric database for U.S. Marine
corps Personnel: summery and statistics. Newton Centre, MA: Geo Centers Inc.,
1996.

Encyclopedia of Surgery. 2015. http://www.surgeryencyclopedia.com/La-


Pa/Ophthalmologic-Surgery.html (accessed 2015).

Engel, D., J. Raczkowsky, and H. Worn. "A safe robot system for craniofacial surgery."
IEEE International Conference on and Automation (ICRA) 2 (2001): 2020-
2024.

Farsi, M., K. Ratcliff, and M. Barbosa. "An introduction to CANopen." Computing &
Control Engineering Journal, Volume 10, 1999.

Fei, B., W. Sing Ng, S. Chauhan, and C. K. Kwoh. "The safety issues of medical
robotics,." Reliability Engineering & System Safety 73, no. 2 (2001): 183-192.

Ghani, Khurshid R., Quoc-Dien Trinh, Jesse Sammon, Wooju Jeong, Ali Dabaja, and
Mani Menon. "Robot-assisted urological surgery: Current status and future
perspectives." Arab Journal of Urology 10 (March 2012): 17-22.

Gheshmi, Siavash. Design and Development of Advanced Surgical Robotic Systems.


Orono: M.Sc. thesis, University of Maine, 2012.

94
Goroll, Allan H., and Albert G. Mulley. Primary Care Medicine: Office Evaluation and
Management of the Adult Patient. 6th. Philadelphia: Lippincott Williams &
Wilkins, 2012.

Graham, J. H. "Research issues in robot safety." IEEE International Conference on


Robotics and Automation (ICRA) 3 (1988): 1854-1855.

Gupta, P.K., P.S. Jensen, and E. de Juan. "Surgical forces and tactile perception during
retinal microsurgery." Medical Image Computing and Computer-Assisted
Intervention, 1999: 1218-1225.

Hendrix, Ron. Robotically assisted eye surgery: A haptic master console. Eindhoven:
Ipskamp Drukkers, TU/e, 2011.

Higson, G. R. Medical Device Safety: The Regulation of Medical Devices for Public
Health and Safety. 1st. London: The Institute of Physics (IoP), 2001.

Ida, Yoshiki, Naohiko Sugita, Takashi Ueta, Yasuhiro Tamaki, Keiji Tanimoto, and
Mamoru Mitsubishi. "Microsurgical robotic system for vitreoretinal surgery."
International Journal of Computer Assisted Radiology and Surgery 7, no. 1
(January 2012): 27-34.

InTouch Health Inc. RP-VITA Remote Presence Robot. 2012.


http://www.intouchhealth.com/products-and-services/products/rp-vita-
robot/ (accessed November 2013).

Intuitive Surgical. Intuitive Surgical - da Vinci Products FAQ. 2013.


http://www.intuitivesurgical.com/products/products_faq.html (accessed
2013).

Jagtap, A. D., and C. N. Riviere. "Applied force during vitreoretinal microsurgery with
handheld instruments." 26th Annual IEEE International Conference of the
Engineering in Medicine and Biology Society. 2004. 2771-2773.

Jensen, P.S., P.K. Gupta, and E. de Juan. "Quantification of microsurgical tactile


perception." Engineering in Medicine and Biology, 21st Annual Conf. and the

95
Fall Meeting of the Biomedical Engineering Society BMES/EMBS Conference.
1999.

Kafader, Urs. "Motion Control for Newbies." 2014.


Katalyst Surgical Inc. Ophthalmic Instruments. 2013.
http://www.katalystsurgical.com/ (accessed 2013).

Kazanzides, P. "Safety design for medical robots." IEEE Annual International


Conference of the Engineering in Medicine and Biology Society, 2009: 7208-
7211.

Kazazides, Peter, and Paul Thienphrapa. "Centralized processing and distributed I/O
for robot control." Technologies for Practical Robot Applications. IEEE
International Conference, 2008. 84-88.

Kirby, Daniel B. "Presentation of Instruments—Instruments for Cataract Surgery."


Trans Am Ophthalmol Soc, 1947: 423-432.

Kohanbash, David. "CAN bus (CANopen & CiA) for Motor Control." 2014.
Kumar, R., et al. "Preliminary experiments in cooperative human/robot force control
for robot assisted microsurgical manipulation." IEEE International Conference
on Robotics and Automation (ICRA) 1 (2000): 610-617.

Lambaudie, E., et al. "Role of robot-assisted laparoscopy in adjuvant surgery for


locally advanced cervical cancer." European Journal of Surgical Oncology
(EJSO) 36, no. 4 (April 2010): 409-413.

Lynch, Patrick J. "Medical Illustrator." Lateral eye and orbit anatomy with nerves.
2006.

Mahpeykar, Navid. Design, Development and Fabrication of an Advanced high


precision Robotic System for Microsurgery. Orono: M.Sc. thesis, University of
Maine, 2013.

Maxon. "Application Notes Collection." 2013.

96
Maxon. CANopen Basic Information. Maxon Corporation, 2008.

Meenink, Thijs. Vitreo-retinal eye surgery robot: sustainable precision, PhD dess.
Eindhoven: Ipskamp Drukkers, 2011.

Meshram, P.M., and R.G. Kanojiya. "Tuning of PID controller using Ziegler-Nichols
method for speed control of DC motor." Advances in Engineering, Science and
Management (ICAESM), 2012.

ORBIS Cyber-Sight. Manual Small Incision Cataract Surgery. New York: ORBIS
International, 2006.

Patel, K. "Robotics the future of surgery." International Journal of Surgery 6, 2008:


441-442.

Pieper, D. The Kinematics of Manipulators Under Computer Control. Stanford


University: Unpublished, 1968.

Pieper, D., and B. Roth. "The Kinematics of Manipulators Under Computer Control."
Proceedings of the Second International Congress on Theory of Machines and
Mechanisms 2 (1969): 159-169.

Pitcher, John D., Jason T. Wilson, Tsu-ChinTsao, Steven D. Schwartz, and Jean-Pierre
Hubschman. "Robotic Eye Surgery: Past, Present, and Future." Computer
Science Systems Biology, Published Online, 2012: doi:10.4172/jcsb.S3-001.

R.Leigh, James. Control theory. 2004.

Rahimy, E., J. Wilson, T. C. Tsao, S. Schwartz, and J. P. Hubschman. "Robot-assisted


intraocular surgery: development of the IRISS and feasibility studies in an
animal model." Eye 27 (August 2013): 972-978.

Robinson, Stephen A. "Cataract Extraction." Department of Surgery, GHSU. Cataract


Extraction. Augusta, GA, 2012.

97
Shahinpoor, M., and S. Gheshmi. Robotic Surgery: Smart Materials, Robotic Structures,
and Artificial Muscles. New York and London: Taylor and Francis Publishers,
PAN Stanford Publishing, 2014.

Tameesh, M. K., R. R. Lakhanpal, and et al. "Retinal vein cannulation with prolonged
infusion of tissue plasminogen activator (t-PA) for the treatment of
experimental retinal vein occlusion in dogs." American Journal of
Ophthalmology 138, no. 5 (2004): 829-839.

Taylor, R. H., et al. "Taming the bull: safety in a precise surgical robot." Fifth
International Conference on Advanced Robotics, ICAR, 1991: 865-870.

Ueta, T., et al. "Robot-Assisted Vitreoretinal Surgery: Development of a Prototype


and Feasibility Studies in an Animal Model." Ophthalmology 116, no. 8
(August 2009): 1538-1543.

98
APPENDIX A: MATLAB CODE FOR FORWARD KINEMATICS

To find symbolic forward kinematic equations:

clear all
clc;
syms Phi Psi Th1 Th2 R Z H
T10=[cos(Phi) sin(Phi) 0 0;
-sin(Phi) cos(Phi) 0 0;
0 0 1 H;
0 0 0 1]
T21=[1 0 0 0;
0 0 -1 R;
0 1 0 0;
0 0 0 1]
T32=[cos(Th1) -sin(Th1) 0 0;
-sin(Th1)*cosd(45) -cos(Th1)*cosd(45) cosd(45) -H;
-sin(Th1)*cosd(45) -cos(Th1)*cosd(45) -cosd(45) H;
0 0 0 1]
T43=[cos(Th2) -sin(Th2) 0 0;
sin(Th2)*cosd(22.5) cos(Th2)*cosd(22.5) -sind(22.5) 0;
sin(Th2)*sind(22.5) cos(Th2)*sind(22.5) cosd(22.5) 0;
0 0 0 1]
T54=[1 0 0 0;
0 cosd(22.5) -sind(22.5) -Z*sind(22.5);
0 sind(22.5) cosd(22.5) Z*cosd(22.5);
0 0 0 1]
T65=[cos(Psi) -sin(Psi) 0 0;
sin(Psi) cos(Psi) 0 0;
0 0 1 0;
0 0 0 1]

T60=T10*T21*T32*T43*T54*T65

To find numerical forward kinematic equations with input arguments in degrees:

function T60=F_Kine(Phi,Th1,Th2,Psi,r,Z)

Phi=deg2rad(Phi);
Th1=deg2rad(Th1);
Th2=deg2rad(Th2);
Psi=deg2rad(Psi);

H=1;
R=H+r;
T10=[cos(Phi) sin(Phi) 0 0;
-sin(Phi) cos(Phi) 0 0;
0 0 1 H;
0 0 0 1];
T21=[1 0 0 0;
0 0 -1 R;
0 1 0 0;

99
0 0 0 1];
T32=[cos(Th1) -sin(Th1) 0 0;
-sin(Th1)*cosd(45) -cos(Th1)*cosd(45) cosd(45) -H;
-sin(Th1)*cosd(45) -cos(Th1)*cosd(45) -cosd(45) H;
0 0 0 1];
T43=[cos(Th2) -sin(Th2) 0 0;
sin(Th2)*cosd(22.5) cos(Th2)*cosd(22.5) -sind(22.5) 0;
sin(Th2)*sind(22.5) cos(Th2)*sind(22.5) cosd(22.5) 0;
0 0 0 1];
T54=[1 0 0 0;
0 cosd(22.5) -sind(22.5) -Z*sind(22.5);
0 sind(22.5) cosd(22.5) Z*cosd(22.5);
0 0 0 1];
T65=[cos(Psi) -sin(Psi) 0 0;
sin(Psi) cos(Psi) 0 0;
0 0 1 0;
0 0 0 1];

T60=T10*T21*T32*T43*T54*T65;

end

100
APPENDIX B: MATLAB CODE FOR INVERSE KINEMATICS

function [Phi Th1 Th2 Psi R Z]=Inv_Kine(T)

r11=T(1,1); r12=T(1,2); r13=T(1,3);


r21=T(2,1); r22=T(2,2); r23=T(2,3);
r31=T(3,1); r32=T(3,2); r33=T(3,3);
Px=T(1,4); Py=T(2,4); Pz=T(3,4);

Phi=atan2(Px-Pz*r13/r33,Py-Pz*r23/r33);
R=sqrt((Px-Pz*r13/r33)^2+(Py-Pz*r23/r33)^2);
Z=Pz/r33;

Rz=[cos(Phi) -sin(Phi) 0 0;
sin(Phi) cos(Phi) 0 0;
0 0 1 0;
0 0 0 1];

%Calculating the rotated {Tool} to obtain {Tool} in {RCM}


Tprime=Rz*T;
r11p=Tprime(1,1); r21p=Tprime(2,1); r31p=Tprime(3,1);
r13p=Tprime(1,3); r23p=Tprime(2,3); r33p=Tprime(3,3);

Th2=2*asin(sqrt((r23p+r33p-1)/0.41421356));

%NOTE: The positive value is chosen


%negative Th2 results in another set of solutions for Th1 and
Psi

A=(0.35355339+0.35355339*cos(Th2));
B=0.38268343*sin(Th2);
C=r13p;

E=(1-sin(Th2/2)^2);
F=0.2705980501*sin(Th2);
D=r23p-0.7071067812*sin(Th2/2)^2;

Sine_Th1=(2*B*D-B*E+E*C)/(2*F*B+A*E);
Cosine_Th1=C/B-A/B*Sine_Th1;

Th1=atan2(Sine_Th1,Cosine_Th1);

Cosine_Psi=((cos(Th1)*cos(Th2)-
0.92387953*sin(Th1)*sin(Th2))*r11p...

+(0.27059805*sin(Th2)+0.65328148*cos(Th1)*sin(Th2)+0.70710678*cos(Th2)*
sin(Th1))*r21p...
+(0.27059805*sin(Th2)-0.65328148*cos(Th1)*sin(Th2)-
0.70710678*cos(Th2)*sin(Th1))*r31p);

Sine_Psi=((0.14644661*sin(Th1)-0.92387953*cos(Th1)*sin(Th2)-
0.85355339*cos(Th2)*sin(Th1))*r11p...
+(2.4142136*sin(0.5*Th1)^2*sin(0.5*Th2)^2-1.0*sin(0.5*Th1)^2-
1.7071068*sin(0.5*Th2)^2-0.65328148*sin(Th1)*sin(Th2)+1.0)*r21p...

101
+(sin(0.5*Th1)^2-
2.4142136*sin(0.5*Th1)^2*sin(0.5*Th2)^2+0.70710678*sin(0.5*Th2)^2+0.653
28148*sin(Th1)*sin(Th2))*r31p);

Psi=atan2(Sine_Psi,Cosine_Psi);

Phi=Phi*180/pi;
Th1=Th1*180/pi;
Th2=Th2*180/pi;
Psi=Psi*180/pi;
disp(' Phi Th1 Th2 Psi r Z')
disp([ Phi Th1 Th2 Psi R Z])

end

102
APPENDIX C: MATLAB CODE TO PLOT RESPONSES OF THE DC MOTOR

SYSTEM

function Arezoo=PIDC(Kp,Ki,Kd,t)

J=0.13*10^-7; % kgm2 %
b=10*10^-8; % Nms/rad %
K=0.01; % Nm/A %
R=27.4; % Ohm %
L=0.399*10^-3; % H %

s = tf([1 0],1);
Gv = K/((L*s + R)*(J*s + b) + K^2);
Ga = Gv/s;

Gv.InputName = 'Voltage';
Gv.OutputName = 'Velocity';
Ga.InputName = 'Voltage';
Ga.OutputName = 'Angle';

% G = [Gv; Ga];
G = ss(Ga);
set(G,'c',[0 1 0; 0 0 1],'d',[0;0],'OutputName',{'Velocity';'Angle'});

% Plotting system properties: response to step input, impulse input


% and bode plot(frequency domain)

% figure(1);
% subplot(2,1,1); step(Ga, 0.03);title('Step
Response','FontSize',10);ylabel('Position
(rad)','FontSize',10);xlabel('Time','FontSize',10);
% subplot(2,1,2); impulse(Ga, 0.03);title('Impulse
Response','FontSize',10);ylabel('Position
(rad)','FontSize',10);xlabel('Time','FontSize',10);
% set(gcf,'Color',[0.85,0.7,1]);
% set(gcf,'InvertHardcopy','off');
% print('-dmeta');

% figure(2);set(gcf,'Color',[0.85,0.7,1])
% subplot(2,1,1); step(Gv, 0.03);title('Step
Response','FontSize',10);ylabel('Velocity
(rad/s)','FontSize',10);xlabel('Time','FontSize',10);
% subplot(2,1,2); impulse(Gv, 0.03);title('Impulse
Response','FontSize',10);ylabel('Velocity
(rad/s)','FontSize',10);xlabel('Time','FontSize',10);
% set(gcf,'InvertHardcopy','off');
% print('-dpng');

% figure(3);
% subplot(2,1,1); bode(Ga);
% subplot(2,1,2); bode(Gv);
% set(gcf,'Color',[0.85,0.7,1]);
% set(gcf,'InvertHardcopy','off');
% print('-dpng');

103
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% % % Proportional controller

% % % Position feedback
Gc = feedback(Ga*Kp,1);% Angle
Gc.InputName = 'Desired Angle';

% figure(5)
% rlocus(Gc)
% set(gcf,'Color',[0.85,0.7,1]);
% set(gcf,'InvertHardcopy','off');
% print('-dpng');

% figure(6);
% step(Gc,0:0.00001:t);title('Step
Response','FontSize',10);ylabel('Position
(rad)','FontSize',10);xlabel('Time','FontSize',10);
% set(gcf,'Color',[0.85,0.7,1]);
% set(gcf,'InvertHardcopy','off');
% print('-dpng');

% % % Velocity feedback
Gc = feedback(Gv*Kp,1);% Velocity
Gc.InputName = 'Desired velocity';

% figure(7); step(Gc,0:0.0005:t);

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

% PID controller

C = tf([Kd Kp Ki],[1 0]);

% Position feedback
Gc = feedback(Ga*C,1);% Angle
Gc.InputName = 'Desired Angle';

figure(8); step(Gc,0:0.00001:t);title('PID Step


Response','FontSize',10);ylabel('Position
(rad)','FontSize',10);xlabel('Time','FontSize',10);
set(gcf,'Color',[0.85,0.7,1]);
set(gcf,'InvertHardcopy','off');
% print('-dpng');

% Velocity feedback
Gc = feedback(Gv*C,1);% Velocity
Gc.InputName = 'Desired velocity';

% figure(9); step(Gc,0:0.0005:t);
end

104
BIOGRAPHY OF THE AUTHOR

Arezoo Ebrahimi was born in Booshehr Port, Iran on January 3rd, 1985. She

attended Bentolhoda Sadr high school in Shiraz. She then studied Industrial

Engineering at Sharif University of Technology and graduated in 2006 with a

Bachelor’s degree in Industrial Engineering.

In September 2013, she started her studies at University of Maine to obtain a

Master of Science degree in Mechanical Engineering. She joined the graduate

program as a teaching and research assistant at Biomedical Engineering and

Advanced Robotics Laboratory.

Arezoo is a candidate for the Master of Science degree in Mechanical Engineering

from the University of Maine in December 2015.

105

You might also like